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 469 records · Page 26

Design Principles for Sparse Matrix Multiplication on the GPU

We implement two novel algorithms for sparse-matrix dense-matrix multiplication (SpMM) on the GPU. Our algorithms expect the sparse input in the popular compressed-sparse-row (CSR) format and thus do not require expensive format conversion. While previous SpMM work concentrates on thread-level parallelism, we additionally focus on latency hiding with instruction-level parallelism and load-balancing. We show, both theoretically and experimentally, that the proposed SpMM is a better fit for the GPU than previous approaches. We identify a key memory access pattern that allows efficient access into both input and output matrices that is crucial to getting excellent performance on SpMM. By combining these two ingredients---(i) merge-based load-balancing and (ii) row-major coalesced memory access---we demonstrate a 4.1x peak speedup and a 31.7% geomean speedup over state-of-the-art SpMM implementations on real-world datasets.

97 MATHEMATICS AND COMPUTING↗

AENET–LAMMPS and AENET–TINKER : Interfaces for accurate and efficient molecular dynamics simulations with machine learning potentials

Machine-learning potentials (MLPs) trained on data from quantum-mechanics based first-principles methods can approach the accuracy of the reference method at a fraction of the computational cost. To facilitate efficient MLP-based molecular dynamics and Monte Carlo simulations, an integration of the MLPs with sampling software is needed. Here, we develop two interfaces that link the atomic energy network (ænet) MLP package with the popular sampling packages TINKER and LAMMPS. The three packages, ænet, TINKER, and LAMMPS, are free and open-source software that enable, in combination, accurate simulations of large and complex systems with low computational cost that scales linearly with the number of atoms. Scaling tests show that the parallel efficiency of the ænet–TINKER interface is nearly optimal but is limited to shared-memory systems. The ænet–LAMMPS interface achieves excellent parallel efficiency on highly parallel distributed memory systems and benefits from the highly optimized neighbor list implemented in LAMMPS. We demonstrate the utility of the two MLP interfaces for two relevant example applications: the investigation of diffusion phenomena in liquid water and the equilibration of nanostructured amorphous battery materials.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

PSelInv—A Distributed Memory Parallel Algorithm for Selected Inversion: The Symmetric Case

We describe an efficient parallel implementation of the selected inversion algorithm for distributed memory computer systems, which we call PSelInv. The PSelInv method computes selected elements of a general sparse matrix Athat can be decomposed as A= LU, where L is lower triangular and U is upper triangular. The implementation described in this article focuses on the case of sparse symmetric matrices. It contains an interface that is compatible with the distributed memory parallel sparse direct factorization SuperLU-DIST. Yet, the underlying data structure and design of PSelInv allows it to be easily combined with other factorization routines, such as PARDISO. We discuss general parallelization strategies such as data and task distribution schemes. Specifically, we describe how to exploit the concurrency exposed by the elimination tree associated with the LU factorization of A. We demonstrate the efficiency and accuracy of PSelInv by presenting several numerical experiments. In particular, we show that PSelInv can run efficiently on more than 4,000 cores for a modestly sized matrix. We also show how PSelInv can be used to accelerate large-scale electronic structure calculations.

97 MATHEMATICS AND COMPUTING↗

Rate 3/4 convolutional coding of 16-PSK - Code design and performance study

Convolutional coding coupled with 16-PSK modulation is investigated for bandwidth efficient transmission. Rate 3/4, small memory codes are found which are optimized in the free-distance sense on the Gaussian channel. These codes provide up to 4.8 dB of coding gain with 32 states over uncoded 8-PSK, a scheme having the same spectral efficiency as the codes described. The performance is compared with earlier findings of Ungerboeck and some recent results on R = 2/3 coded 8-PSK. In addition, results of a channel transmission study to assess the performance of the four-state code on the band-limited nonlinear channel are presented, and it is found that performance of the coded scheme degrades comparably with uncoded 8-PSK, i.e., coding gain is roughly preserved.

Wilson, S. G.↗

In situ Parallel Training of Analog Neural Network Using Electrochemical Random-Access Memory

In-memory computing based on non-volatile resistive memory can significantly improve the energy efficiency of artificial neural networks. However, accurate in situ training has been challenging due to the nonlinear and stochastic switching of the resistive memory elements. One promising analog memory is the electrochemical random-access memory (ECRAM), also known as the redox transistor. Its low write currents and linear switching properties across hundreds of analog states enable accurate and massively parallel updates of a full crossbar array, which yield rapid and energy-efficient training. While simulations predict that ECRAM based neural networks achieve high training accuracy at significantly higher energy efficiency than digital implementations, these predictions have not been experimentally achieved. In this work, we train a 3 × 3 array of ECRAM devices that learns to discriminate several elementary logic gates (AND, OR, NAND). We record the evolution of the network’s synaptic weights during parallel in situ (on-line) training, with outer product updates. Due to linear and reproducible device switching characteristics, our crossbar simulations not only accurately simulate the epochs to convergence, but also quantitatively capture the evolution of weights in individual devices. The implementation of the first in situ parallel training together with strong agreement with simulation results provides a significant advance toward developing ECRAM into larger crossbar arrays for artificial neural network accelerators, which could enable orders of magnitude improvements in energy efficiency of deep neural networks.

97 MATHEMATICS AND COMPUTING↗

A three dimensional multigrid Reynolds-averaged Navier-Stokes solver for unstructured meshes

A three-dimensional unstructured mesh Reynolds averaged Navier-Stokes solver is described. Turbulence is simulated using a single field-equation model. Computational overheads are minimized through the use of a single edge-based data-structure, and efficient multigrid solution technique, and the use of multi-tasking on shared memory multi-processors. The accuracy and efficiency of the code are evaluated by computing two-dimensional flows in three dimensions and comparing with results from a previously validated two-dimensional code which employs the same solution algorithm. The feasibility of computing three-dimensional flows on grids of several million points in less than two hours of wall clock time is demonstrated.

Mavriplis, D. J.↗

Ultrafast high-endurance memory based on sliding ferroelectrics

The persistence of voltage-switchable collective electronic phenomena down to the atomic scale has extensive implications for area- and energy-efficient electronics, especially in emerging nonvolatile memory technology. We investigate the performance of a ferroelectric field-effect transistor (FeFET) based on sliding ferroelectricity in bilayer boron nitride at room temperature. Sliding ferroelectricity represents a different form of atomically thin two-dimensional (2D) ferroelectrics, characterized by the switching of out-of-plane polarization through interlayer sliding motion. We examined the FeFET device employing monolayer graphene as the channel layer, which demonstrated ultrafast switching speeds on the nanosecond scale and high endurance exceeding 10 11 switching cycles, comparable to state-of-the-art FeFET devices. These characteristics highlight the potential of 2D sliding ferroelectrics for inspiring next-generation nonvolatile memory technology.

Science & Technology - Other Topics↗

Scalable Work Stealing

Irregular and dynamic parallel applications pose significant challenges to achieving scalable performance on large-scale multicore clusters. These applications often require ongoing, dynamic load balancing in order to maintain efficiency. While effective at small scale, centralized load balancing schemes quickly become a bottleneck on large-scale clusters. Work stealing is a popular approach to distributed dynamic load balancing; however its performance on large-scale clusters is not well understood. Prior work on work stealing has largely focused on shared memory machines. In this work we investigate the design and scalability of work stealing on modern distributed memory systems. We demonstrate high efficiency and low overhead when scaling to 8,192 processors for three benchmark codes: a producer-consumer benchmark, the unbalanced tree search benchmark, and a multiresolution analysis kernel.

99 GENERAL AND MISCELLANEOUS↗

Developement of 3D Vertically Integrated Pattern Recognition Associative Memory (VIPRAM)

Many next-generation physics experiments will be characterized by the collection of large quantities of data, taken in rapid succession, from which scientists will have to unravel the underlying physical processes. In most cases, large backgrounds will overwhelm the physics signal. Since the quantity of data that can be stored for later analysis is limited, real-time event selection is imperative to retain the interesting events while rejecting the background. Scaling of current technologies is unlikely to satisfy the scientific needs of future projects, so investments in transformational new technologies need to be made. For example, future particle physics experiments looking for rare processes will have to address the demanding challenges of fast pattern recognition in triggering as detector hit density becomes significantly higher due to the high luminosity required to produce the rare processes. In this proposal, we intend to develop hardware-based technology that significantly advances the state-of-the-art for fast pattern recognition within and outside HEP using the 3D vertical integration technology that has emerged recently in industry. The ultimate physics reach of the LHC experiments will crucially depend on the tracking trigger's ability to help discriminate between interesting rare events and the background. Hardware-based pattern recognition for fast triggering on particle tracks has been successfully used in high-energy physics experiments for some time. The CDF Silicon Vertex Trigger (SVT) at the Fermilab Tevatron is an excellent example. The method used there, developed in the 1990's, is based on algorithms that use a massively parallel associative memory architecture to identify patterns efficiently at high speed. However, due to much higher occupancy and event rates at the LHC, and the fact that the LHC detectors have a much larger number of channels in their tracking detectors, there is an enormous challenge in implementing pattern recognition for a track trigger, requiring about three orders of magnitude more associative memory patterns than what was used in the original CDF SVT. Significant improvement in the architecture of associative memory structures is needed to run fast pattern recognition algorithms of this scale. We are proposing the development of 3D integrated circuit technology as a way to implement new associative memory structures for fast pattern recognition applications. Adding a 'third' dimension to the signal processing chain, as compared to the two-dimensional nature of printed circuit boards, Field Programmable Gate Arrays (FPGAs), etc., opens up the possibility for new architectures that could dramatically enhance pattern recognition capability. We are currently performing preliminary design work to demonstrate the feasibility of this approach. In this proposal, we seek to develop the design and perform the ASIC engineering necessary to realize a prototype device. While our focus here is on the Energy Frontier (e.g. the LHC), the approach may have applications in experiments in the Intensity Frontier and the Cosmic Frontier as well as other scientific and medical projects. In fact, the technique that we are proposing is very generic and could have wide applications far beyond track trigger, both within and outside HEP.

43 PARTICLE ACCELERATORS↗

A new concept of vertically integrated pattern recognition associative memory

Hardware-based pattern recognition for fast triggering on particle tracks has been successfully used in high-energy physics experiments for some time. The CDF Silicon Vertex Trigger (SVT) at the Fermilab Tevatron is an excellent example. The method used there, developed in the 1990's, is based on algorithms that use a massively parallel associative memory architecture to identify patterns efficiently at high speed. However, due to much higher occupancy and event rates at the LHC, and the fact that the LHC detectors have a much larger number of channels in their tracking detectors, there is an enormous challenge in implementing fast pattern recognition for a track trigger, requiring about three orders of magnitude more associative memory patterns than what was used in the original CDF SVT. Scaling of current technologies is unlikely to satisfy the scientific needs of the future, and investments in transformational new technologies need to be made. In this paper, we will discuss a new concept of using the emerging 3D vertical integration technology to significantly advance the state-of-the-art for fast pattern recognition within and outside HEP. A generic R and D proposal based on this new concept, with a few institutions involved, has recently been submitted to DOE with the goal to design and perform the ASIC engineering necessary to realize a prototype device. The progress of this R and D project will be reported in the future. Here we will only focus on the concept of this new approach.

43 PARTICLE ACCELERATORS↗

Field-free switching of perpendicular magnetization in a ferrimagnetic insulator with spin reorientation transition

Writing magnetic bits through spin-orbit torque (SOT) switching is promising for fast and efficient magnetic random-access memory devices. While SOT switching of out-of-plane (OOP) magnetized states requires lateral symmetry breaking, in-plane (IP) magnetized states suffer from low storage density. Here, we demonstrate a field-free switching scheme using a 5-nanometer europium iron garnet film grown with a (110) orientation that shows a spin reorientation transition from OOP to IP above room temperature. This scheme combines the benefits of high-density storage in the OOP states at room temperature and the efficient field-free SOT switching in the IP states at elevated temperatures. While conventional switching of OOP bits faces the dilemma that high OOP anisotropy is required to improve bit stability and low OOP anisotropy is required to lower switching current density, this scheme disentangles this interdependence, allowing for low switching currents to be possible without sacrificing the bit stability, offering opportunities for future memory devices.

Science & Technology - Other Topics↗

Parallel performance investigations of an unstructured mesh Navier-Stokes solver

A Reynolds-averaged Navier-Stokes solver based on unstructured mesh techniques for analysis of high-lift configurations is described. The method makes use of an agglomeration multigrid solver for convergence acceleration. Implicit line-smoothing is employed to relieve the stiffness associated with highly stretched meshes. A GMRES technique is also implemented to speed convergence at the expense of additional memory usage. The solver is cache efficient and fully vectorizable, and is parallelized using a two-level hybrid MPI-OpenMP implementation suitable for shared and/or distributed memory architectures, as well as clusters of shared memory machines. Convergence and scalability results are illustrated for various high-lift cases.

Mavriplis, Dimitri J.↗

LDRD Abbreviated report: High-Order General-Discrete-Ordinates Method Enabling Efficient Deterministic Transport in Hydrodynamic Simulations

Deterministic transport simulations for national-security and energy applications often operate in high-dimensional phase-space, where accuracy and cost both become major challenges. A common numerical artifact in such problems is the “ray-effect,” which appears as unphysical streaks. Beyond misinterpretation, these artifacts can contaminate tightly coupled physics, such as fluid dynamics, radiation-hydrodynamics, and laser-plasma interactions, eroding the predictive capability of entire multiphysics workflows. Our objective was to make high-dimension studies practical on modern hardware while mitigating the ray-effect without relying on prohibitively expensive sampling approaches such as Monte Carlo methods. We developed the Generic Discretization Library (GenDiL), a Graphics Processing Unit (GPU)-first framework that uses high-order Discontinuous Galerkin (DG) methods and matrix-free algorithms to reduce memory usage and improve computational efficiency, critical for phase-space simulations. GenDiL supports phase-space adaptivity in both mesh size and polynomial order (hp-adaptivity) to place resolution only where it is needed. A central capability is Local Dimensional Refinement (LDR), which couples lower-dimension continuum models to higher-dimension kinetic models through stable and conservative interfaces, so that high-fidelity physics is applied only in regions where it is essential. Building on the GenDiL framework, we developed the General SN (GSN) family of algorithms as a true generalization of the polar SN approach (discrete ordinates, often denoted SN). Rather than tying discrete ordinates to a specific polar change of coordinates, GSN formulates transport on an arbitrary change of coordinates chosen to reduce ray-effect. We studied two complementary variants: an analytic variant, where the coordinate map is prescribed in advance by a closed-form function; and a data-driven variant, where a quantity of interest, such as the net flux, guides the coordinate system. GenDiL provides the library infrastructure for efficient GPU execution, but the GSN concept is algorithmic and independent of any one library. Across representative high-dimension tests, including non-symmetric solutions, both variants delivered strong ray-effect mitigation at practical cost, moving four- to six-dimensional analysis toward repeatable, routine studies.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Dynamical nonlinear memory capacitance in biomimetic membranes

Two-terminal memory elements, or memelements, capable of co-locating signal processing and memory via history-dependent reconfigurability at the nanoscale are vital for next-generation computing materials striving to match the brain’s efficiency and flexible cognitive capabilities. While memory resistors, or memristors, have been widely reported, other types of memelements remain underexplored or undiscovered. Here we report the first example of a volatile, voltage-controlled memcapacitor in which capacitive memory arises from reversible and hysteretic geometrical changes in a lipid bilayer that mimics the composition and structure of biomembranes. We demonstrate that the nonlinear dynamics and memory are governed by two implicitly-coupled, voltage-dependent state variables—membrane radius and thickness. Further, our system is capable of tuneable signal processing and learning via synapse-like, short-term capacitive plasticity. These findings will accelerate the development of low-energy, biomolecular neuromorphic memelements, which, in turn, could also serve as models to study capacitive memory and signal processing in neuronal membranes.

59 BASIC BIOLOGICAL SCIENCES↗

Final Report: Correctness Tools for Petascale Computing

In the course of developing parallel programs for leadership computing systems, subtle programming errors often arise that are extremely difficult to diagnose without tools. To meet this challenge, University of Maryland, the University of Wisconsin—Madison, and Rice University worked to develop lightweight tools to help code developers pinpoint a variety of program correctness errors that plague parallel scientific codes. The aim of this project was to develop software tools that help diagnose program errors including memory leaks, memory access errors, round-off errors, and data races. Research at Rice University focused on developing algorithms and data structures to support efficient monitoring of multithreaded programs for memory access errors and data races. This is a final report about research and development work at Rice University as part of this project.

97 MATHEMATICS AND COMPUTING↗

Performance Optimization Methods for a Memory-Bound, Unstructured-Grid CFD Application on Massively Parallel GPU Platforms

Computational performance of the FUN3D unstructured-grid computational fluid dynamics (CFD) application on massively parallel GPU environments is memory-bound and highly dependent upon efficient reads from and atomic updates to the irregular cell-, edge-, and node-based data structures. In this talk, we present recent efforts into optimizing select performance-critical kernels on NVIDIA Tesla V100 and A100 GPUs and AMD CDNA MI100 GPUs. A novel use of L2 cache residency controls and asynchronous loads into on-chip shared memory are explored on the A100 GPU for the sparse iterative solver, which is dominated by mixed-precision, sparse matrix vector multiplication. Demonstrations show that these methods improve global memory bandwidth utilization by 13.5% on the A100 GPU. Several techniques are also presented that use registers and/or shared memory to facilitate array transposition and aggregation which combine to reduce the frequency and increase the cache efficiency of floating-point atomic updates to the irregular data structures. These methods are demonstrated to improve the kernel throughput by nearly 500% on select kernels on the AMD MI100 over atomic updates directly to global memory. Overall, both V100 and A100 GPUs outperformed the MI100 GPU on kernels dominated by double-precision atomic updates; however, the techniques demonstrated here reduced the performance gap and improved the MI100 performance.

GPU CPU unstructured CFD memory↗