Search NASA⌕ Search

SEARCH · Search NASA

Results for “memory efficiency”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 649 records · Page 36

UPC++ v1.0 Programmer’s Guide, Revision 2021.9.0

UPC++ is a C++ library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate constituent process having access to local memory as it would in C++. PGAS additionally provides one-sided Remote Memory Access (RMA) to a global address space, which is allocated in shared segments that are distributed over the processes. UPC++ also features Remote Procedure Call (RPC) communication, making it easy to move computation to operate on data that resides on remote processes. In UPC++, all communication operations are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all communication operations are asynchronous by default, to enable programmers to write code that scales well even on hundreds of thousands of cores.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Dispatch Manager for NEML2 Constitutive Model Calculations Embedded in MOOSE

This report describes the extended capabilities of the NEML2 constitutive modeling library, including a flexible and efficient work dispatching system designed to leverage both CPU and GPU resources. This enhancement addresses one of the primary computational challenges in large-scale simulations: the ability to distribute and execute batches of material model evaluations across heterogeneous computing devices. The new dispatch system introduces a modular set of dispatcher and scheduler classes that coordinate the flow of data and execution between devices. The dispatcher is responsible for efficiently packaging work, managing device-specific memory operations, and synchronizing results. This modularity allows for extensibility, making it straightforward to integrate additional computing backends in the future. From an implementation standpoint, the dispatcher system interfaces seamlessly with NEML2's existing models. They handle device-aware tensor operations, optimize memory transfers, and support asynchronous execution when applicable. This design ensures that batches of material points can be evaluated concurrently, substantially improving throughput compared to previous single-device or serial implementations. These improvements not only enhance the raw performance of NEML2 but also improve its usability in multiscale and high-fidelity simulations, where the simultaneous evaluation of large material point batches is critical. Benchmarks included in the report demonstrate the system’s scalability, highlighting its effectiveness when leveraging modern GPU architectures.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

UPC++ Programmer’s Guide, v1.0-2018.9.0

UPC++ is a C++11 library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate constituent process having access to local memory as it would in C++. However, PGAS also provides access to a global address space, which is allocated in shared segments that are distributed over the processes. UPC++ provides numerous methods for accessing and using global memory. In UPC++, all operations that access remote memory are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all remote-memory access operations are by default asynchronous, to enable programmers to write code that scales well even on hundreds of thousands of cores.

97 MATHEMATICS AND COMPUTING↗

UPC++ v1.0 Programmer’s Guide, Revision 2020.3.0

UPC++ is a C++11 library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate constituent process having access to local memory as it would in C++. However, PGAS also provides access to a global address space, which is allocated in shared segments that are distributed over the processes. UPC++ provides numerous methods for accessing and using global memory. In UPC++, all operations that access remote memory are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all remote-memory access operations are by default asynchronous, to enable programmers to write code that scales well even on hundreds of thousands of cores.

97 MATHEMATICS AND COMPUTING↗

Plasma simulation using the massively parallel processor

Two dimensional electrostatic simulation codes using the particle-in-cell model are developed on the Massively Parallel Processor (MPP). The conventional plasma simulation procedure that computes electric fields at particle positions by means of a gridded system is found inefficient on the MPP. The MPP simulation code is thus based on the gridless system in which particles are assigned to processing elements and electric fields are computed directly via Discrete Fourier Transform. Currently, the gridless model on the MPP in two dimensions is about nine times slower that the gridded system on the CRAY X-MP without considering I/O time. However, the gridless system on the MPP can be improved by incorporating a faster I/O between the staging memory and Array Unit and a more efficient procedure for taking floating point sums over processing elements. The initial results suggest that the parallel processors have the potential for performing large scale plasma simulations.

Lin, C. S.↗

Storage and computationally efficient permutations of factorized covariance and square-root information arrays

A unified method to permute vector stored Upper triangular Diagonal factorized covariance and vector stored upper triangular Square Root Information arrays is presented. The method involves cyclic permutation of the rows and columns of the arrays and retriangularization with fast (slow) Givens rotations (reflections). Minimal computation is performed, and a one dimensional scratch array is required. To make the method efficient for large arrays on a virtual memory machine, computations are arranged so as to avoid expensive paging faults. This method is potentially important for processing large volumes of radio metric data in the Deep Space Network.

Muellerschoen, R. J.↗

Parallel computing for probabilistic fatigue analysis

This paper presents the results of Phase I research to investigate the most effective parallel processing software strategies and hardware configurations for probabilistic structural analysis. We investigate the efficiency of both shared and distributed-memory architectures via a probabilistic fatigue life analysis problem. We also present a parallel programming approach, the virtual shared-memory paradigm, that is applicable across both types of hardware. Using this approach, problems can be solved on a variety of parallel configurations, including networks of single or multiprocessor workstations. We conclude that it is possible to effectively parallelize probabilistic fatigue analysis codes; however, special strategies will be needed to achieve large-scale parallelism to keep large number of processors busy and to treat problems with the large memory requirements encountered in practice. We also conclude that distributed-memory architecture is preferable to shared-memory for achieving large scale parallelism; however, in the future, the currently emerging hybrid-memory architectures will likely be optimal.

Sues, Robert H.↗

Communications and information research: Improved space link performance via concatenated forward error correction coding

With the development of new advanced instruments for remote sensing applications, sensor data will be generated at a rate that not only requires increased onboard processing and storage capability, but imposes demands on the space to ground communication link and ground data management-communication system. Data compression and error control codes provide viable means to alleviate these demands. Two types of data compression have been studied by many researchers in the area of information theory: a lossless technique that guarantees full reconstruction of the data, and a lossy technique which generally gives higher data compaction ratio but incurs some distortion in the reconstructed data. To satisfy the many science disciplines which NASA supports, lossless data compression becomes a primary focus for the technology development. While transmitting the data obtained by any lossless data compression, it is very important to use some error-control code. For a long time, convolutional codes have been widely used in satellite telecommunications. To more efficiently transform the data obtained by the Rice algorithm, it is required to meet the a posteriori probability (APP) for each decoded bit. A relevant algorithm for this purpose has been proposed which minimizes the bit error probability in the decoding linear block and convolutional codes and meets the APP for each decoded bit. However, recent results on iterative decoding of 'Turbo codes', turn conventional wisdom on its head and suggest fundamentally new techniques. During the past several months of this research, the following approaches have been developed: (1) a new lossless data compression algorithm, which is much better than the extended Rice algorithm for various types of sensor data, (2) a new approach to determine the generalized Hamming weights of the algebraic-geometric codes defined by a large class of curves in high-dimensional spaces, (3) some efficient improved geometric Goppa codes for disk memory systems and high-speed mass memory systems, and (4) a tree based approach for data compression using dynamic programming.

Rao, T. R. N.↗

Re-Design and Beat Testing of the Man-Machine Integration Design and Analysis System: MIDAS

The Man-machine Design and Analysis System (MIDAS) is a human factors design and analysis system that combines human cognitive models with 3D CAD models and rapid prototyping and simulation techniques. MIDAS allows designers to ask 'what if' types of questions early in concept exploration and development prior to actual hardware development. The system outputs predictions of operator workload, situational awareness and system performance as well as graphical visualization of the cockpit designs interacting with models of the human in a mission scenario. Recently, MIDAS was re-designed to enhance functionality and usability. The goals driving the redesign include more efficient processing, GUI interface, advances in the memory structures, implementation of external vision models and audition. These changes were detailed in an earlier paper. Two Beta test sites with diverse applications have been chosen. One Beta test site is investigating the development of a new airframe and its interaction with the air traffic management system. The second Beta test effort will investigate 3D auditory cueing in conjunction with traditional visual cueing strategies including panel-mounted and heads-up displays. The progress and lessons learned on each of these projects will be discussed.

Shively, R. Jay↗

Exploring Machine Learning Techniques For Dynamic Modeling on Future Exascale Systems

Future exascale systems must be optimized for both power and performance at scale in order to achieve DOE’s goal of a sustained petaflop within 20 Megawatts by 2022 [1]. Massive parallelism of the future systems combined with complex memory hierarchies will form a barrier to efficient application and architecture design. These challenges are exacerbated with emerging complex architectures such as GPGPUs and Intel Xeon Phi as parallelism increases orders of magnitude and system power consumption can easily triple or quadruple. Therefore, we need techniques that can reduce the search space for optimization, isolate power-performance bottlenecks, identify root causes for software/hardware inefficiency, and effectively direct runtime scheduling.

Song, Shuaiwen↗

Exploiting Thread Parallelism for Ocean Modeling on Cray XC Supercomputers

The incorporation of increasing core counts in modern processors used to build state-of-the-art supercomputers is driving application development towards exploitation of thread parallelism, in addition to distributed memory parallelism, with the goal of delivering efficient high-performance codes. In this work we describe the exploitation of threading and our experiences with it with respect to a real-world ocean modeling application code, MPAS-Ocean. We present detailed performance analysis and comparisons of various approaches and configurations for threading on the Cray XC series supercomputers.

97 MATHEMATICS AND COMPUTING↗

A Communication-Optimal Framework for Contracting Distributed Tensors

Tensor contractions are extremely compute intensive generalized matrix multiplication operations encountered in many computational science fields, such as quantum chemistry and nuclear physics. Unlike distributed matrix multiplication, which has been extensively studied, limited work has been done in understanding distributed tensor contractions. In this paper, we characterize distributed tensor contraction algorithms on torus networks. We develop a framework with three fundamental communication operators to generate communication-efficient contraction algorithms for arbitrary tensor contractions. We show that for a given amount of memory per processor, our framework is communication optimal for all tensor contractions. We demonstrate performance and scalability of our framework on up to 262,144 cores of BG/Q supercomputer using five tensor contraction examples.

communication efficiency↗

Dynamic implicit 3D adaptive mesh refinement for non-equilibrium radiation diffusion

The time dependent non-equilibrium radiation diffusion equations are important for solving the transport of energy through radiation in optically thick regimes and find applications in several fields including astrophysics and inertial confinement fusion. The associated initial boundary value problems that are encountered often exhibit a wide range of scales in space and time and are extremely challenging to solve. To efficiently and accurately simulate these systems we describe our research on combining techniques that will also find use more broadly for long term time integration of nonlinear multi-physics systems: implicit time integration for efficient long term time integration of stiff multi-physics systems, local control theory based step size control to minimize the required global number of time steps while controlling accuracy, dynamic 3D adaptive mesh refinement (AMR) to minimize memory and computational costs, Jacobian Free Newton–Krylov methods on AMR grids for efficient nonlinear solution, and optimal multilevel preconditioner components that provide level independent solver convergence.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Optimizing Distributed Training on Frontier for Large Language Models

Large language models (LLMs) have demonstrated remarkable success as foundational models, benefiting various downstream applications through fine-tuning. Loss scaling studies have demonstrated the superior performance of larger LLMs compared to their smaller counterparts. Nevertheless, training LLMs with billions of parameters poses significant challenges and requires considerable computational resources. For example, training a one trillion parameter GPT-style model on 20 trillion tokens requires a staggering 120 million exaflops. This research explores efficient distributed training strategies to extract this computation from Frontier, the world's first exascale supercomputer. We enable and investigate various model and data parallel training techniques, such as tensor parallelism, pipeline parallelism, and sharded data parallelism, to facilitate training a trillion-parameter model on Frontier. We empirically assess these techniques and their associated parameters to determine their impact on memory footprint, communication latency, and GPU's computational efficiency. We analyze the complex interplay among these techniques and find a strategy to combine them to achieve high throughput through hyperparameter tuning. We have identified efficient strategies for training large LLMs of varying sizes through empirical analysis and hyperparameter tuning. For 22 Billion, 175 Billion, and 1 Trillion parameters, we achieved GPU throughputs of 38.38%, 36.14%, and 31.96%, respectively. For the training of the 175 Billion parameter model and the 1 Trillion parameter model, we achieved 100% weak scaling efficiency on 1024 and 3072 Mi250X GPUs, respectively. We also achieved strong scaling efficiencies of 89% and 87% for these two models. We trained these models only tens of iterations instead of training till completion.

Yin, Junqi↗

NASA Tech Briefs, June 2009

Topics covered include: Device for Measuring Low Flow Speed in a Duct, Measuring Thermal Conductivity of a Small Insulation Sample, Alignment Jig for the Precise Measurement of THz Radiation, Autoignition Chamber for Remote Testing of Pyrotechnic Devices, Microwave Power Combiners for Signals of Arbitrary Amplitude, Synthetic Foveal Imaging Technology, Airborne Antenna System for Minimum-Cycle-Slip GPS Reception, Improved Starting Materials for Back-Illuminated Imagers, Multi-Modulator for Bandwidth-Efficient Communication, Some Improvements in Utilization of Flash Memory Devices, GPS/MEMS IMU/Microprocessor Board for Navigation, T/R Multi-Chip MMIC Modules for 150 GHz, Pneumatic Haptic Interfaces, Device Acquires and Retains Rock or Ice Samples, Cryogenic Feedthrough Test Rig, Improved Assembly for Gas Shielding During Welding or Brazing, Two-Step Plasma Process for Cleaning Indium Bonding Bumps, Tool for Crimping Flexible Circuit Leads, Yb14MnSb11 as a High-Efficiency Thermoelectric Material, Polyimide-Foam/Aerogel Composites for Thermal Insulation, Converting CSV Files to RKSML Files, Service Management Database for DSN Equipment, Chemochromic Hydrogen Leak Detectors, Compatibility of Segments of Thermoelectric Generators, Complementary Barrier Infrared Detector, JPL Greenland Moulin Exploration Probe, Ultra-Lightweight Self-Deployable Nanocomposite Structure for Habitat Applications, and Room-Temperature Ionic Liquids for Electrochemical Capacitors.

Source record↗

TriC: Distributed-memory Triangle Counting by Exploiting the Graph Structure

Graph analytics has emerged as an important tool in the analysis of large scale data from diverse application domains such as social networks, cyber security and bioinformatics. Counting the number of triangles in a graph is a fundamental kernel with several applications such as detecting the community structure of a graph or in identifying important vertices in a graph. The ubiquity of massive datasets is driving the need to scale graph analytics on parallel systems. However, numerous challenges exist in efficiently parallelizing graph algorithms, especially on distributed-memory systems. Irregular memory accesses and communication patterns, low computation to communication ratios, and the need for frequent synchronization are some of the leading challenges. In this paper, we present TriC, our distributed-memory implementation of triangle counting in graphs using the Message Passing Interface (MPI), as a submission to the 2020 GraphChallenge competition. Using a set of synthetic and real-world inputs from the challenge, we demonstrate a speedup of up to 90x relative to previous work on 32 processor-cores of a NERSC Cori node. We also provide details from distributed runs with up to8192 processes along with strong scaling results. The observations presented in this work provide an understanding of the system-level bottlenecks at scale that specifically impact sparse-irregular workloads and will therefore benefit other efforts to parallelize graph algorithms.

Halappanavar, Mahantesh↗

UltraLiM: In-Memory Boolean Logic Architecture Using UltraRAM

Conventional computing architectures encounter ‘von Neumann’ and ‘memory wall’ bottlenecks which arise due to the back-and-forth data movement between the physically separate memory and processing units and the speed mismatch between them, respectively. These bottlenecks hurt both energy efficiency and the throughput of computing systems. To address these challenges, in-memory computing architectures have emerged as a promising alternative. They reduce the need for frequent data movement by executing different computing tasks inside the memory system. Here, we present UltraLiM, a logic-in-memory architecture using the UltraRAM-based memory system. UltraRAM holds the promise of developing a ‘universal memory’, overcoming the limitations of charge-based memories thanks to their non-volatile behavior with lower operating voltage. This work presents an in-memory computing architecture that integrates an UltraRAM-based memory array with a custom-designed peripheral circuitry. With this architecture, we can perform various in-memory Boolean logic operations (such as NOT, NAND, NOR, and XOR) in a single cycle. Leveraging the separate read-write paths in the UltraRAM-based memory array, we optimize read operations without encountering design conflicts. This optimization enhances the sense margin, enabling the use of simpler peripheral circuitry for in-memory logic operations.

Alam, Shamiul [University of Tennessee, Knoxville ↗

Exploring Manycore Multinode Systems for Irregular Applications with FPGA Prototyping

We present a prototype of a multi-core architecture implemented on FPGA, designed to enable efficient execution of irregular applications on distributed shared memory machines, while maintaining high performance on regular workloads. The architecture is composed of off-the-shelf soft-core cores, local interconnection and memory interface, integrated with custom components that optimize it for irregular applications. It relies on three key elements: a global address space, multithreading, and fine-grained synchronization. Global addresses are scrambled to reduce the formation of network hot-spots, while the latency of the transactions is covered by integrating an hardware scheduler within the custom load/store buffers to take advantage from the availability of multiple executions threads, increasing the efficiency in a transparent way to the application. We evaluated a dual node system irregular kernels showing scalability in the number of cores and threads.

FPGA↗