Search NASA⌕ Search

SEARCH · Search NASA

Results for “near memory computing”

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 199 records · Page 11

GEOS Atmospheric Model: Challenges at Exascale

The Goddard Earth Observing System (GEOS) model at NASA's Global Modeling and Assimilation Office (GMAO) is used to simulate the multi-scale variability of the Earth's weather and climate, and is used primarily to assimilate conventional and satellite-based observations for weather forecasting and reanalysis. In addition, assimilations coupled to an ocean model are used for longer-term forecasting (e.g., El Nino) on seasonal to interannual times-scales. The GMAO's research activities, including system development, focus on numerous time and space scales, as detailed on the GMAO website, where they are tabbed under five major themes: Weather Analysis and Prediction; Seasonal-Decadal Analysis and Prediction; Reanalysis; Global Mesoscale Modeling, and Observing System Science. A brief description of the GEOS systems can also be found at the GMAO website. GEOS executes as a collection of earth system components connected through the Earth System Modeling Framework (ESMF). The ESMF layer is supplemented with the MAPL (Modeling, Analysis, and Prediction Layer) software toolkit developed at the GMAO, which facilitates the organization of the computational components into a hierarchical architecture. GEOS systems run in parallel using a horizontal decomposition of the Earth's sphere into processing elements (PEs). Communication between PEs is primarily through a message passing framework, using the message passing interface (MPI), and through explicit use of node-level shared memory access via the SHMEM (Symmetric Hierarchical Memory access) protocol. Production GEOS weather prediction systems currently run at 12.5-kilometer horizontal resolution with 72 vertical levels decomposed into PEs associated with 5,400 MPI processes. Research GEOS systems run at resolutions as fine as 1.5 kilometers globally using as many as 30,000 MPI processes. Looking forward, these systems can be expected to see a 2 times increase in horizontal resolution every two to three years, as well as less frequent increases in vertical resolution. Coupling these resolution changes with increases in complexity, the computational demands on the GEOS production and research systems should easily increase 100-fold over the next five years. Currently, our 12.5 kilometer weather prediction system narrowly meets the time-to-solution demands of a near-real-time production system. Work is now in progress to take advantage of a hybrid MPI-OpenMP parallelism strategy, in an attempt to achieve a modest two-fold speed-up to accommodate an immediate demand due to increased scientific complexity and an increase in vertical resolution. Pursuing demands that require a 10- to 100-fold increases or more, however, would require a detailed exploration of the computational profile of GEOS, as well as targeted solutions using more advanced high-performance computing technologies. Increased computing demands of 100-fold will be required within five years based on anticipated changes in the GEOS production systems, increases of 1000-fold can be anticipated over the next ten years.

ESMF↗

Towards Scalable Deep Learning via I/O Analysis and Optimization

Deep learning systems have been growing in prominence as a way to automatically characterize objects, trends, and anomalies. Given the importance of deep learning systems, researchers have been investigating techniques to optimize such systems. An area of particular interest has been using large supercomputing systems to quickly generate effective deep learning networks: a phase often referred to as “training” of the deep learning neural network. As we scale existing deep learning frameworks—such as Caffe—on these large supercomputing systems, we notice that the parallelism can help improve the computation tremendously, leaving data I/O as the major bottleneck limiting the overall system scalability. In this paper, we first present a detailed analysis of the performance bottlenecks of Caffe on large supercomputing systems. Our analysis shows that the I/O subsystem of Caffe—LMDB—relies on memory-mapped I/O to access its database, which can be highly inefficient on large-scale systems because of its interaction with the process scheduling system and the network-based parallel filesystem. Based on this analysis, we then present LMDBIO, our optimized I/O plugin for Caffe that takes into account the data access pattern of Caffe in order to vastly improve I/O performance. Our experimental results show that LMDBIO can improve the overall execution time of Caffe by nearly 20-fold in some cases.

Pumma, Sarunya↗

Large-scale quasi-Newton trust-region methods with low-dimensional linear equality constraints

We propose two limited-memory BFGS (L-BFGS) trust-region methods for large-scale optimization with linear equality constraints. Here, the methods are intended for problems where the number of equality constraints is small. By exploiting the structure of the quasi-Newton compact representation, both proposed methods solve the trust-region subproblems nearly exactly, even for large problems. We derive theoretical global convergence results of the proposed algorithms, and compare their numerical effectiveness and performance on a variety of large-scale problems.

97 MATHEMATICS AND COMPUTING↗

Data, Photographs, Videos, and Information for the Niwot Ridge Subalpine Forest (US-NR1) AmeriFlux site

This data package contains data and information about the operation of the Niwot Ridge Subalpine Forest AmeriFlux site (US-NR1) between Nov 1998 to the present (2020). This data archive supplements the primary 30-min data storage for the US-NR1 data (i.e., https://doi.org/10.17190/AMF/1246088) by providing the following: (i) five-minute statistics (means, variances, covariances) of all data measured by the data system between Nov 1998 and September 2020 in netCDF format, (ii) CSV data files saved within the memory of the CR23X data loggers (as well as an archive of the data logger programs), (iii) an archive of previous 30-min ASCII data versions of the US-NR1 AmeriFlux data and information related to each data release (a replica of what can be found at http://urquell.colorado.edu/data_ameriflux/), (iv) a web calendar (in HTML format) documenting activity at the site (a replica of http://urquell.colorado.edu/calendar/), (v) photos (over 15,000) and video taken at the site between years 2001 and present day (2020), and (vi) several auxiliary datasets, primary related to trees near the site, soil moisture and soil temperature, and subcanopy radiation data. The data package is setup so that the web calendar, photos, and electronic logbook can be easily accessed on a local computer using a web browser. The provided data files are in either netCDF, CSV, ASCII, or MATLAB format. To obtain a better understanding about the archive, please start by reading the PDF: README_ESS_DIVE_USNR1_readme_first.pdf.

54 ENVIRONMENTAL SCIENCES↗

UPC++: A High-Performance Communication Framework for Asynchronous Computation

© 2019 IEEE UPC++ is a C++ library that supports high-performance computation via an asynchronous communication framework. This paper describes a new incarnation that differs substantially from its predecessor, and we discuss the reasons for our design decisions. We present new design features, including future-based asynchrony management, distributed objects, and generalized Remote Procedure Call (RPC). We show microbenchmark performance results demonstrating that one-sided Remote Memory Access (RMA) in UPC++ is competitive with MPI-3 RMA; on a Cray XC40 UPC++ delivers up to a 25% improvement in the latency of blocking RMA put, and up to a 33% bandwidth improvement in an RMA throughput test. We showcase the benefits of UPC++ with irregular applications through a pair of application motifs, a distributed hash table and a sparse solver component. Our distributed hash table in UPC++ delivers near-linear weak scaling up to 34816 cores of a Cray XC40. Our UPC++ implementation of the sparse solver component shows robust strong scaling up to 2048 cores, where it outperforms variants communicating using MPI by up to 3.1x. UPC++ encourages the use of aggressive asynchrony in low-overhead RMA and RPC, improving programmer productivity and delivering high performance in irregular applications.

Bachan, John↗

Near-memory data reorganization engine

A memory subsystem package is provided that has processing logic for data reorganization within the memory subsystem package. The processing logic is adapted to reorganize data stored within the memory subsystem package. In some embodiments, the memory subsystem package includes memory units, a memory interconnect, and a data reorganization engine ("DRE"). The data reorganization engine includes a stream interconnect and DRE units including a control processor and a load-store unit. The control processor is adapted to execute instructions to control a data reorganization. The load-store unit is adapted to process data move commands received from the control processor via the stream interconnect for loading data from a load memory address of a memory unit and storing data to a store memory address of a memory unit.

97 MATHEMATICS AND COMPUTING↗

Automatic off-body overset adaptive Cartesian mesh method based on an octree approach

This paper describes a method for generating adaptive structured Cartesian grids within a near-body/off-body mesh partitioning framework for the flow simulation around complex geometries. The off-body Cartesian mesh generation derives from an octree structure, assuming each octree leaf node defines a structured Cartesian block. This enables one to take into account the large scale discrepancies in terms of resolution between the different bodies involved in the simulation, with minimum memory requirements. Two different conversions from the octree to Cartesian grids are proposed: the first one generates Adaptive Mesh Refinement (AMR) type grid systems, and the second one generates abutting or minimally overlapping Cartesian grid set. We also introduce an algorithm to control the number of points at each adaptation, that automatically determines relevant values of the refinement indicator driving the grid refinement and coarsening. An application to a wing tip vortex computation assesses the capability of the method to capture accurately the flow features.

97 MATHEMATICS AND COMPUTING↗

Cusps enable line attractors for neural computation

Here, line attractors in neuronal networks have been suggested to be the basis of many brain functions, such as working memory, oculomotor control, head movement, locomotion, and sensory processing. In this paper, we make the connection between line attractors and pulse gating in feed-forward neuronal networks. In this context, because of their neutral stability along a one-dimensional manifold, line attractors are associated with a time-translational invariance that allows graded information to be propagated from one neuronal population to the next. To understand how pulse-gating manifests itself in a high-dimensional, nonlinear, feedforward integrate-and-fire network, we use a Fokker-Planck approach to analyze system dynamics. We make a connection between pulse-gated propagation in the Fokker-Planck and population-averaged mean-field (firing rate) models, and then identify an approximate line attractor in state space as the essential structure underlying graded information propagation. An analysis of the line attractor shows that it consists of three fixed points: a central saddle with an unstable manifold along the line and stable manifolds orthogonal to the line, which is surrounded on either side by stable fixed points. Along the manifold defined by the fixed points, slow dynamics give rise to a ghost. We show that this line attractor arises at a cusp catastrophe, where a fold bifurcation develops as a function of synaptic noise; and that the ghost dynamics near the fold of the cusp underly the robustness of the line attractor. Understanding the dynamical aspects of this cusp catastrophe allows us to show how line attractors can persist in biologically realistic neuronal networks and how the interplay of pulse gating, synaptic coupling, and neuronal stochasticity can be used to enable attracting one-dimensional manifolds and, thus, dynamically control the processing of graded information.

59 BASIC BIOLOGICAL SCIENCES↗

Analyzing inference workloads for spatiotemporal modeling

Ensuring power grid resiliency, forecasting climate conditions, and optimization of transportation infrastructure are some of the many application areas where data is collected in both space and time. Spatiotemporal modeling is about modeling those patterns for forecasting future trends and carrying out critical decision-making by leveraging machine learning/deep learning. Once trained offline, field deployment of trained models for near real-time inference could be challenging because performance can vary significantly depending on the environment, available compute resources and tolerance to ambiguity in results. Users deploying spatiotemporal models for solving complex problems can benefit from analytical studies considering a plethora of system adaptations to understand the associated performance-quality trade-offs. To facilitate the co-design of next-generation hardware architectures for field deployment of trained models, it is critical to characterize the workloads of these deep learning (DL) applications during inference and assess their computational patterns at different levels of the execution stack. In this paper, we develop several variants of deep learning applications that use spatiotemporal data from dynamical systems. We study the associated computational patterns for inference workloads at different levels, considering relevant models (Long short-term Memory, Convolutional Neural Network and Spatio-Temporal Graph Convolution Network), DL frameworks (Tensorflow and PyTorch), precision (FP16, FP32, AMP, INT16 and INT8), inference runtime (ONNX and AI Template), post-training quantization (TensorRT) and platforms (Nvidia DGX A100 and Sambanova SN10 RDU). Overall, our findings indicate that although there is potential in mixed-precision models and post-training quantization for spatiotemporal modeling, extracting efficiency from contemporary GPU systems might be challenging. Instead, co-designing custom accelerators by leveraging optimized High Level Synthesis frameworks (such as SODA High-Level Synthesizer for customized FPGA/ASIC targets) can make workload-specific adjustments to enhance the efficiency.

97 MATHEMATICS AND COMPUTING↗

Toward memory-efficient melt pool monitoring: a classification framework using event-based imaging and sparse sensing technique

Vision sensors like CMOS and CCD cameras are often used for in-process monitoring of melt pools in laser-based additive and welding processes, but they require transferring large amounts of data and computational processing resources. Event-based neuromorphic imagery, on the other hand, detects only the change in pixel intensity, thus potentially reducing the data amount and latency. With an event imager, this study develops a framework for melt pool condition classification, including image construction, time scale selection, optimal pixel selection, and sparse classification, to achieve a highly memory-efficient scheme. These are based on sparse sensing techniques with singular value decomposition (SVD) and QR pivoting, the two fundamental matrix transformations for linear dimensionality reduction. The framework is then validated by classifying a controlled experiment by exciting various mode shapes of liquid gallium pools of varying depths (3, 6, and 8 mm). At 200 pixels, the classifier can reach overall accuracy of 75%, while at 2000 pixels (0.013% of the total possible pixels), the accuracy is nearly 90% (89.86%). At the same number of pixels, random selection can only achieve 46% and 67%, respectively. The memory savings of the sparsely sampled event data compared to a conventional imager is about 500 times. In addition to performance, implementation and limitations of the framework are also discussed.

42 ENGINEERING↗

DEPEND: A simulation-based environment for system level dependability analysis

The design and evaluation of highly reliable computer systems is a complex issue. Designers mostly develop such systems based on prior knowledge and experience and occasionally from analytical evaluations of simplified designs. A simulation-based environment called DEPEND which is especially geared for the design and evaluation of fault-tolerant architectures is presented. DEPEND is unique in that it exploits the properties of object-oriented programming to provide a flexible framework with which a user can rapidly model and evaluate various fault-tolerant systems. The key features of the DEPEND environment are described, and its capabilities are illustrated with a detailed analysis of a real design. In particular, DEPEND is used to simulate the Unix based Tandem Integrity fault-tolerance and evaluate how well it handles near-coincident errors caused by correlated and latent faults. Issues such as memory scrubbing, re-integration policies, and workload dependent repair times which affect how the system handles near-coincident errors are also evaluated. Issues such as the method used by DEPEND to simulate error latency and the time acceleration technique that provides enormous simulation speed up are also discussed. Unlike any other simulation-based dependability studies, the use of these approaches and the accuracy of the simulation model are validated by comparing the results of the simulations, with measurements obtained from fault injection experiments conducted on a production Tandem Integrity machine.

Goswami, Kumar↗

Martensitic Phase Transition in Complex NiTi-Based Shape Memory Alloys from First-Principles Calculations

Recent rapid progresses in physics theory and computational power have made it possible to accurately predict the phase transitions and martensitic transition temperatures (MTTs) in shape memory alloys (SMAs) from first principles. However, previously theory and calculations [1-4] were applied only to study highly ordered stoichiometric binary alloys such as NiTi, PdTi and NiHf. Here we report on our recent first-principles investigations [5,6] on Ni 0.5 Ti 0.5-x Hf x and Pd x Ni 0.5-x Ti 0.5 ternaries and off-stoichiometric NiTi, and the predicted martensitic phase transitions in these complex SMAs are in good agreement with experimental findings. In particular, the calculated MTTs for all these compositions are within 100 K compared with the corresponding measured data, and our results also reveal the origin of the striking asymmetry in MTT of the off-stoichiometric NiTi near equiatomic compositions. We will address various techniques to overcome the difficulty encountered in studying ternaries and off-stoichiometic binaries associated with disorder and/or much lowered symmetry. Our theoretical approach is expected to be a broadly applicable and predictive theory for designing complex SMAs with desirable properties. References: [1] J. B. Haskins, A. E. Thompson, and J. W. Lawson, Phys. Rev B 94 , 214110 (2016). [2] J. B. Haskins and J. W. Lawson, J. App. Phys. 121 , 205103 (2017). [3] J. B. Haskins, H. Malmir, S. J. Honrao, L. A. Sandoval, and J. W. Lawson, Acta Materialia 212 , 116872 (2017). [4] Z. Wu, J. W. Lawson, and O. Benafan, Phys. Rev. B 106 , L140102 (2022). [5] Z. Wu, H. Malmir, O. Benafan, and J. W. Lawson, Acta Materialia 261 , 119362 (2023). [6] Z. Wu, J. W. Lawson, and O. Benafan, Phys. Rev. B 108 , L140103 (2023).

Zhigang Wu↗

The SGI/Cray T3E: Experiences and Insights

The NASA Goddard Space Flight Center is home to the fifth most powerful supercomputer in the world, a 1024 processor SGI/Cray T3E-600. The original 512 processor system was placed at Goddard in March, 1997 as part of a cooperative agreement between the High Performance Computing and Communications Program's Earth and Space Sciences Project (ESS) and SGI/Cray Research. The goal of this system is to facilitate achievement of the Project milestones of 10, 50 and 100 GFLOPS sustained performance on selected Earth and space science application codes. The additional 512 processors were purchased in March, 1998 by the NASA Earth Science Enterprise for the NASA Seasonal to Interannual Prediction Project (NSIPP). These two "halves" still operate as a single system, and must satisfy the unique requirements of both aforementioned groups, as well as guest researchers from the Earth, space, microgravity, manned space flight and aeronautics communities. Few large scalable parallel systems are configured for capability computing, so models are hard to find. This unique environment has created a challenging system administration task, and has yielded some insights into the supercomputing needs of the various NASA Enterprises, as well as insights into the strengths and weaknesses of the T3E architecture and software. The T3E is a distributed memory system in which the processing elements (PE's) are connected by a low latency, high bandwidth bidirectional 3-D torus. Due to the focus on high speed communication between PE's, the T3E requires PE's to be allocated contiguously per job. Further, jobs will only execute on the user specified number of PE's and PE timesharing is possible but impractical. With a highly varied job mix in both size and runtime of jobs, the resulting scenario is PE fragmentation and an inability to achieve near 100% utilization. SGI/Cray has provided several scheduling and configuration tools to minimize the impact of fragmentation. These tools include PScheD (the political scheduler), GRM (the global resource manager) and NQE (the Network Queuing Environment). Features and impact of these tools will be discussed, as will resulting performance and utilization data. As a distributed memory system, the T3E is designed to be programmed through explicit message passing. Consequently, certain assumptions related to code design are made by the operating system (UNICOS/mk) and its scheduling tools. With the exception of HPF, which does run on the T3E, however poorly, alternative programming styles have the potential to impact the T3E in unexpected and undesirable ways. Several examples will be presented (preceeded with the disclaimer, "Don't try this at home! Violators will be prosecuted!")

Bernard, Lisa Hamet↗

Data Reduction Techniques for Simulation, Visualization and Data Analysis

Data reduction is increasingly being applied to scientific data for numerical simulations, scientific visualizations, and data analyses. It is most often used to lower I/O and storage costs, and sometimes to lower in-memory data size as well. With this work, we consider five categories of data reduction techniques based on their information loss: 1) truly lossless, 2) near lossless, 3) lossy, 4) mesh reduction, and 5) derived representations. We then survey available techniques in each of these categories, summarize their properties from a practical point of view, and discuss relative merits within a category. We believe, in total, this work will enable simulation scientists and visualization/data analysis scientists to decide which data reduction techniques will be most helpful for their needs.

97 MATHEMATICS AND COMPUTING↗

Parallel String Graph Construction and Transitive Reduction for De Novo Genome Assembly

One of the most computationally intensive tasks in computational biology is de novo genome assembly, the decoding of the sequence of an unknown genome from redundant and erroneous short sequences. A common assembly paradigm identifies overlapping sequences, simplifies their layout, and creates consensus. Despite many algorithms developed in the literature, the efficient assembly of large genomes is still an open problem. In this work, we introduce new distributed-memory parallel algorithms for overlap detection and layout simplification steps of de novo genome assembly, and implement them in the diBELLA 2D pipeline. Our distributed memory algorithms for both overlap detection and layout simplification are based on linear-algebra operations over semirings using 2D distributed sparse matrices. Our layout step consists of performing a transitive reduction from the overlap graph to a string graph. We provide a detailed communication analysis of the main stages of our new algorithms. diBELLA 2D achieves near linear scaling with over 80% parallel efficiency for the human genome, reducing the runtime for overlap detection by 1.2-1.3× for the human genome and 1.5-1.9× for C.elegans compared to the state-of-the-art. Our transitive reduction algorithm outperforms an existing distributed-memory implementation by 10.5-13.3× for the human genome and 18-29× for the C. elegans. Our work paves the way for efficient de novo assembly of large genomes using long reads in distributed memory.

59 BASIC BIOLOGICAL SCIENCES↗

Affinity-aware checkpoint restart

Current checkpointing techniques employed to overcome faults for HPC applications result in inferior application performance after restart from a checkpoint for a number of applications. This is due to a lack of page and core affinity awareness of the checkpoint/restart (C/R) mechanism, i.e., application tasks originally pinned to cores may be restarted on different cores, and in case of non-uniform memory architectures (NUMA), quite common today, memory pages associated with tasks on a NUMA node may be associated with a different NUMA node after restart. Here, this work contributes a novel design technique for C/R mechanisms to preserve task-to-core maps and NUMA node specific page affinities across restarts. Experimental results with BLCR, a C/R mechanism, enhanced with affinity awareness demonstrate significant performance benefits of 37%-73% for the NAS Parallel Benchmark codes and 6-12% for NAMD with negligible overheads instead of up to nearly four times longer an execution times without affinity-aware restarts on 16 cores.

97 MATHEMATICS AND COMPUTING↗

Density Matrix Quantum Circuit Simulation via the BSP Machine on Modern GPU Clusters

As quantum computers evolve, simulations of quantum programs on classical computers will be essential in validating quantum algorithms, understanding the effect of system noise, and design applications for future quantum computers. In this paper, we propose a novel multi-GPU programming model called MG-BSP that constructs a virtual BSP machine on top of modern multi-GPU platforms, and tweak the programming model to build a multi-GPU density matrix quantum simulator. We propose and evaluated a new formulation minimizing communication and prove that the transformation conserves original semantics when noise is introduced. We build the tool-chain to support quantum assembly open standard, synthesize testing quantum circuit, and enable ultra-deep quantum simulation. We evaluated our design on four state-of-the-art multi-GPU platforms including the latest DGX-1 and DGX-2 systems. We demonstrate simulation of 1 million gates in 94 minutes, far deeper circuits than has been demonstrated in prior work. A roofline-model analysis show that we have reached near-optimal performance under memory bound.

Li, Ang↗

Memory embedded non-intrusive reduced order modeling of non-ergodic flows

Generating a digital twin of any complex system requires modeling and computational approaches that are efficient, accurate, and modular. Traditional reduced order modeling techniques are targeted at only the first two, but the novel nonintrusive approach we present in this study is an attempt at taking all three into account effectively compared to their traditional counterparts. Based on dimensionality reduction using proper orthogonal decomposition (POD), we introduce a long short-term memory neural network architecture together with a principal interval decomposition (PID) framework as an enabler to account for localized modal deformation. As an effective partitioning tool for breaking the Kolmogorov barrier, our PID framework, therefore, can be considered a key element in the accurate reduced order modeling of convective flows. Our applications for convection-dominated systems governed by Burgers, Navier-Stokes, and Boussinesq equations demonstrate that the proposed approach yields significantly more accurate predictions than the POD-Galerkin method and could be a key enabler toward near real-time predictions of unsteady flows.

42 ENGINEERING↗