Engineering Papers⌕ Search

SEARCH · Engineering Papers

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 739 records · Page 41
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Ground and excited state gradients with end-to-end differentiable semiempirical quantum chemistry

Accurate and efficient gradients of molecular energy with respect to nuclear degrees of freedom are essential for geometry optimization and molecular dynamics, including simulations that go beyond the Born–Oppenheimer regime. A common approach involves deriving analytical formulas for new electronic structure methods, which is often conceptually difficult and requires tedious coding. Here, we implement analytical, semi-numerical, and automatic differentiation (AD)-based gradient pathways for semiempirical Hamiltonian models in the PYSEQM software package, leveraging both graphics processing unit (GPU) and central processing unit (CPU) architectures. We further extend these capabilities to excited states calculated using the configuration interaction singles and time-dependent Hartree–Fock ansätze. We benchmark wall time, peak memory usage, and accuracy across three molecular families of varying chemical complexity, including systems of up to a thousand atoms. For ground-state simulations, analytical and AD gradients achieve near-identical GPU runtimes, while semi-numerical gradients are slower on GPU but remain competitive on CPU. For excited states, both analytical and custom AD approaches using implicit differentiation show similar performance and low memory requirements, whereas gradients with full AD are memory-limited. AD gradients match analytical ones in accuracy across all tested systems, aided by a quaternion-based diatomic frame rotation for two-center quantities that ensures smooth energy surfaces. Overall, automatic differentiation emerges as a practical alternative to analytical gradients in semiempirical quantum chemistry, offering high accuracy while allowing seamless integration in AI-driven workflows and popular packages, such as PyTorch and JAX. Our results provide actionable guidance for selecting optimal gradient strategies in large-scale ground- and excited-state molecular dynamics simulations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Matrix-free preconditioning for high-order H (curl) discretizations

The greater arithmetic intensity of high-order finite element discretizations makes them attractive for implementation on next-generation hardware, but assembly of high-order finite element operators as matrices is prohibitively expensive. As a result, the development of general algebraic solvers for such operators has been an open research challenge. Fast matrix-free application of high-order operators has received significant attention in the literature in the context of Poisson-type problems, but preconditioners and solvers for inverting more general operators are not very well-developed. In this paper, we consider the problem of preconditioning a definite Maxwell operator at high polynomial order without assembling a matrix. We show that given efficient preconditioners for high-order H 1 finite element problems on the same mesh, efficient H(curl) preconditioners can be constructed in an auxiliary space framework. We demonstrate the resulting preconditioners in a practical setting with tensor-product basis functions on an unstructured mesh of quadrilaterals. Overall, our approach uses a sparsified H 1 solver constructed on a low-order mesh of the nodal points of the underlying high-order space, and we show that the resulting H(curl) preconditioner is effective at very high polynomial orders for two-dimensional model problems with complicated geometry, varying piecewise constant coefficients, and curved elements. The resulting preconditioner scales with nearly optimal O(p d+1 ) floating point operation count and optimal O(p d ) memory transfer requirements, outperforming existing Maxwell preconditioners in the high-order regime.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

The Adjoint Petrov–Galerkin method for non-linear model reduction

Here, we formulate a new projection-based reduced-order modeling technique for non-linear dynamical systems. The proposed technique, which we refer to as the Adjoint Petrov–Galerkin (APG) method, is derived by decomposing the generalized coordinates of a dynamical system into a resolved coarse-scale set and an unresolved fine-scale set. A Markovian finite memory assumption within the Mori–Zwanzig formalism is then used to develop a reduced-order representation of the coarse scales. This procedure leads to a closed reduced-order model that displays commonalities with the adjoint stabilization method used in finite elements. The formulation is shown to be equivalent to a Petrov–Galerkin method with a non-linear, time-varying test basis, thus sharing some similarities with the Least-Squares Petrov–Galerkin method. Theoretical analysis examining a priori error bounds and computational cost is presented. Numerical experiments on the compressible Navier–Stokes equations demonstrate that the proposed method can lead to improvements in numerical accuracy, robustness, and computational efficiency over the Galerkin method on problems of practical interest. Improvements in numerical accuracy and computational efficiency over the Least-Squares Petrov–Galerkin method are observed in most cases.

42 ENGINEERING↗
DOE OSTI2021

Reduced Order Model of Transactive Bidding Loads

Transactive energy (TE) has been identified to provide better grid efficiency and reliability by market-based transactive exchanges between energy producers and energy consumers. Simulations of TE systems are crucial to evaluate the benefits and impacts of different transactive mechanisms. However, such simulations can be time consuming due to the information exchange between various participants and complex co-simulation environments. In this paper, we develop a reduced order model to speed up the simulation of transactive systems in TE simulation platform (TESP) while achieving very low error between the reduced order and full model results. Specifically, the developed reduced order model consists of an aggregate responsive load agent which utilizes two Recurrent Neural Networks (RNNs) with Long Short-Term Memory units (LSTMs) to enable transactive elements to collectively participate in the TE system. The proposed aggregate responsive load (ARL) agent is able to produce similar transactive behaviors to the full simulation model while achieving significant simulation time reduction. Finally, we also show that the developed model enables generalization of simulation results across different dates and across different number of loads included in the simulations.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2020

Erratum to: Soft- and reactive landing of ions onto surfaces: concepts and applications

Soft- and reactive landing of mass-selected ions is gaining attention as a promising approach for the precisely-controlled preparation of materials on surfaces that are not amenable to deposition using conventional methods. A broad range of ionization sources and mass filters are available that make ion soft-landing a versatile tool for surface modification using beams of hyperthermal (<100?eV) ions. The ability to select the mass-to-charge ratio of the ion, its kinetic energy and charge state, along with precise control of the size, shape, and position of the ion beam on the deposition target distinguishes ion soft landing from other surface modification techniques. Soft- and reactive landing have been used to prepare interfaces for practical applications as well as precisely-defined model surfaces for fundamental investigations in chemistry, physics, and materials science. For instance, soft- and reactive landing have been applied to study the surface chemistry of ions isolated in the gas-phase, prepare arrays of proteins for high-throughput biological screening, produce novel carbon-based and polymer materials, enrich the secondary structure of peptides and the chirality of organic molecules, immobilize electrochemically-active proteins and organometallics on electrodes, create thin films of complex molecules, and immobilize catalytically active organometallics as well as ligated metal clusters. In addition, soft landing has enabled investigation of the size-dependent behavior of bare metal clusters in the critical subnanometer size regime where chemical and physical properties do not scale predictably with size. The morphology, aggregation, and immobilization of larger bare metal nanoparticles, which are directly relevant to the design of catalysts as well as improved memory and electronic devices, have also been studied using ion soft landing. This review article begins in section 1 with a brief introduction to the existing applications of ion soft- and reactive landing. Section 2 provides an overview of the ionization sources and mass filters that have been used to date for soft landing of mass-selected ions. A discussion of the competing processes that occur during ion deposition as well as the types of ions and surfaces that have been investigated follows in section 3. Section 4 discusses the physical phenomena that occur during and after ion soft landing, including retention and reduction of ionic charge along with factors that impact the efficiency of ion deposition. The influence of soft landing on the secondary structure and biological activity of complex ions is addressed in section 5. Lastly, an overview of the structure and mobility as well as the catalytic, optical, magnetic, and redox properties of bare ionic clusters and nanoparticles deposited onto surfaces is presented in section 6. © 2015 Wiley Periodicals, Inc. Mass Spec Rev 35:439–479, 2016.

Johnson, Grant E.↗
DOE OSTI2021

Dimensionality reduction of the many-body problem using coupled-cluster subsystem flow equations: classical and quantum computing perspective

We discuss reduced-scaling strategies employing recently introduced sub-system embedding sub-algebras coupled-cluster formalism (SES-CC) to describe many-body systems. These strategies utilize properties of the SES-CC formulations where the equations describing certain classes of sub- systems can be integrated into a computational flows composed coupled eigenvalue problems of reduced dimensionality. Additionally, these flows can be defined at the level of the CC Ansatz defined by selected classes of cluster amplitudes, which define the wave function ”memory” of possible partitionings of the many-body system into constituent sub-systems. One of the possible ways of solving these coupled problems is through implementing procedures, where the information is passed between the sub-systems in a self-consistent manner. As a special case, we consider local flow formulations where the so-called local character of correlation effects can be closely related to properties of sub-system embedding sub-algebras employing localized molecular basis. We also generalize flow equations to the time domain and to downfolding methods utilizing double exponential unitary CC Ansatz (DUCC), where reduced dimensionality of constituent sub-problems offer a possibility of efficient utilization of limited quantum resources in modeling realistic systems.

Electron correlation, quantum chemistry, quantum c↗
DOE OSTI2020

Cutting the Deployment Costs of Physics-Based MPC in Buildings by Simulation-Based Imitation Learning

It has been shown that model predictive control (MPC) is a promising solution for energy-efficient building operations. However, the deployment of MPC in a large portion of the building stock has not been possible partially because of high installation costs. Every building is unique and requires a tailored MPC solution. The best performing solutions are often based on physics-based modeling, which is, however, computationally expensive and requires dedicated software. A promising direction that tackles this problem is to train a neural network-based optimal control policy to imitate the behavior of physics-based MPC from the simulation data generated offline. The neural networks give control actions that closely approximate those produced by physics-based MPC, but with a fraction of the computational and memory requirements and without the need for licensed software. The main advantage of the proposed approach stems from simple evaluation at execution time, leading to low computational footprints and easy deployment on embedded HW platforms. In the case study, we present the energy savings potential of physics-based MPC applied to an office building in Belgium. We demonstrate how neural network approximators can be used to cut the implementation and maintenance costs of MPC deployment without compromising performance. We also critically assess the presented approach by pointing out the remaining challenges and open research questions.

Drgona, Jan↗
DOE OSTI2020

Compare linear-system solver and preconditioner stacks with emphasis on GPU performance and propose phase-2 NGP solver development pathway

The goal of the ExaWind project is to enable predictive simulations of wind farms comprised of many megawatt-scale turbines situated in complex terrain. Predictive simulations will require computational fluid dynamics (CFD) simulations for which the mesh resolves the geometry of the turbines and captures the rotation and large deflections of blades. Whereas such simulations for a single turbine are arguably petascale class, multi-turbine wind farm simulations will require exascale-class resources. The primary physics codes in the ExaWind project are Nalu-Wind, which is an unstructured-grid solver for the acoustically incompressible Navier-Stokes equations, and OpenFAST, which is a whole-turbine simulation code. The Nalu-Wind model consists of the mass-continuity Poisson-type equation for pressure and a momentum equation for the velocity. For such modeling approaches, simulation times are dominated by linear-system setup and solution for the continuity and momentum systems. For the ExaWind challenge problem, the moving meshes greatly affect overall solver costs as reinitialization of matrices and recomputation of preconditioners is required at every time step. In this report we evaluated GPU-performance baselines for the linear solvers in the Trilinos and hypre solver stacks using two representative Nalu-Wind simulations: an atmospheric boundary layer precursor simulation on a structured mesh, and a fixed-wing simulation using unstructured overset meshes. Both strong-scaling and weak-scaling experiments were conducted on the OLCF supercomputer Summit and similar proxy clusters. We focused on the performance of multi-threaded Gauss-Seidel and two-stage Gauss-Seidel that are extensions of classical Gauss-Seidel; of one-reduce GMRES, a communication-reducing variant of the Krylov GMRES; and algebraic multigrid methods that incorporate the afore-mentioned methods. The team has established that AMG methods are capable of solving linear systems arising from the fixed-wing overset meshes on CPU, a critical intermediate result for ExaWind FY20 Q3 and Q4 milestones. For the fixed-wing strong-scaling study (model with 3M grid-points), the team identified that Nalu-Wind simulations with the new Trilinos and hypre solvers scale to modest GPU counts, maintaining above 70% efficiency up to 6 GPUs. However, there still remain significant bottlenecks to performance: matrix assembly (hypre), AMG setup (hypre and Trilinos) In the weak-scaling experiments (going from 0.4M to 211M gridpoints), it's shown that the solver apply phases are faster on GPUs, but that Nalu-Wind simulation times grow, primarily due to the multigrid-setup process. Finally, based on the report outcomes, we propose a linear solver path-forward for the remainder of the ExaWind project. Near term, the NREL team will continue their work on GPU-based linear-system assembly. They will also investigate how the use of alternatives to the NVIDIA UVM (unified virtual memory) paradigm affects performance. Longer term, the NREL team will evaluate algorithmic performance on other types of accelerators and merge their improvements back to the main hypre repository branch. Near term, the Trilinos team will address performance bottlenecks identified in this milestone, such as implementing a GPU-based segregated momentum solve and reusing matrix graphs across linear-system assembly phases. Longer term, the Trilinos team will do detailed analysis and optimization of multigrid setup.

17 WIND ENERGY↗
DOE OSTI2022

Multiplicative Shot-Noise: A New Route to Stability of Plastic Networks

Fluctuations of synaptic weights, among many other physical, biological, and ecological quantities, are driven by coincident events of two “parent” processes. Here we propose a multiplicative shot-noise model that can capture the behaviors of a broad range of such natural phenomena, and analytically derive an approximation that accurately predicts its statistics. We apply our results to study the effects of a multiplicative synaptic plasticity rule that was recently extracted from measurements in physiological conditions. Using mean-field theory analysis and network simulations, we investigate how this rule shapes the connectivity and dynamics of recurrent spiking neural networks. The multiplicative plasticity rule is shown to support efficient learning of input stimuli, and it gives a stable, unimodal synaptic-weight distribution with a large fraction of strong synapses. The strong synapses remain stable over long times but do not “run away.” Our results suggest that the multiplicative shot-noise offers a new route to understand the tradeoff between flexibility and stability in neural circuits and other dynamic networks.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Emergent ferroelectricity in subnanometer binary oxide films on silicon

The critical size limit of voltage-switchable electric dipoles has extensive implications for energy-efficient electronics, underlying the importance of ferroelectric order stabilized at reduced dimensionality. Here, we report on the thickness-dependent antiferroelectric-to-ferroelectric phase transition in zirconium dioxide (ZrO 2 ) thin films on silicon. The emergent ferroelectricity and hysteretic polarization switching in ultrathin ZrO 2 , conventionally a paraelectric material, notably persists down to a film thickness of 5 angstroms, the fluorite-structure unit-cell size. This approach to exploit three-dimensional centrosymmetric materials deposited down to the two-dimensional thickness limit, particularly within this model fluorite-structure system that possesses unconventional ferroelectric size effects, offers substantial promise for electronics, demonstrated by proof-of-principle atomic-scale nonvolatile ferroelectric memory on silicon. Additionally, it is also indicative of hidden electronic phenomena that are achievable across a wide class of simple binary materials.

42 ENGINEERING↗
DOE OSTI2025

Tree tensor network hierarchical equations of motion based on time-dependent variational principle for efficient open quantum dynamics in structured thermal environments

In this work, we introduce an efficient method, TTN-HEOM, for exactly calculating the open quantum dynamics for driven quantum systems interacting with highly structured bosonic baths by combining the tree tensor network (TTN) decomposition scheme with the bexcitonic generalization of the numerically exact hierarchical equations of motion (HEOM). The method yields a series of quantum master equations for all core tensors in the TTN that efficiently and accurately capture the open quantum dynamics for non-Markovian environments to all orders in the system–bath interaction. These master equations are constructed based on the time-dependent Dirac–Frenkel variational principle, which isolates the optimal dynamics for the core tensors given the TTN ansatz. The dynamics converges to the HEOM when increasing the rank of the core tensors, a limit in which the TTN ansatz becomes exact. We introduce TENSO, tensor equations for non-Markovian structured open systems, as a general-purpose Python code to propagate the TTN-HEOM dynamics. We implement three general propagators for the coupled master equations: two fixed-rank methods that require a constant memory footprint during the dynamics and one adaptive-rank method with a variable memory footprint controlled by the target level of computational error. We exemplify the utility of these methods by simulating a two-level system coupled to a structured bath containing one Drude–Lorentz component and eight Brownian oscillators, which is beyond what can presently be computed using the standard HEOM. Our results show that the TTN-HEOM is capable of simulating both dephasing and relaxation dynamics of driven quantum systems interacting with structured baths, even those of chemical complexity, with an affordable computational cost.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

ProvLight: Efficient Workflow Provenance Capture on the Edge-to-Cloud Continuum

Modern scientific workflows require hybrid infrastructures combining numerous decentralized resources on the IoT/Edge interconnected to Cloud/HPC systems (aka the Computing Continuum) to enable their optimized execution. Understanding and optimizing the performance of such complex Edge-to-Cloud workflows is challenging. Capturing the provenance of key performance indicators, with their related data and processes, may assist in understanding and optimizing workflow executions. However, the capture overhead can be prohibitive, particularly in resource-constrained devices, such as the ones on the IoT/Edge.To address this challenge, based on a performance analysis of existing systems, we propose ProvLight, a tool to enable efficient provenance capture on the IoT/Edge. We leverage simplified data models, data compression and grouping, and lightweight transmission protocols to reduce overheads. We further integrate ProvLight into the E2Clab framework to enable workflow provenance capture across the Edge-to-Cloud Continuum. This integration makes E2Clab a promising platform for the performance optimization of applications through reproducible experiments.We validate ProvLight at a large scale with synthetic workloads on 64 real-life IoT/Edge devices in the FIT IoT LAB testbed. Evaluations show that ProvLight outperforms state-of-the-art systems like ProvLake and DfAnalyzer in resource-constrained devices. ProvLight is 26—37x faster to capture and transmit provenance data; uses 5—7x less CPU; 2x less memory; transmits 2x less data; and consumes 2—2.5x less energy. ProvLight [1] and E2Clab [2] are available as open-source tools.

Rosendo, Daniel↗
DOE OSTI2021

Simple, low-cost and accurate data-driven geophysical forecasting with learned kernels

Modelling geophysical processes as low-dimensional dynamical systems and regressing their vector field from data is a promising approach for learning emulators of such systems. We show that when the kernel of these emulators is also learned from data (using kernel flows, a variant of cross-validation), then the resulting data-driven models are not only faster than equation-based models but are easier to train than neural networks such as the long short-term memory neural network. In addition, they are also more accurate and predictive than the latter. When trained on geophysical observational data, for example the weekly averaged global sea-surface temperature, considerable gains are also observed by the proposed technique in comparison with classical partial differential equation-based models in terms of forecast computational cost and accuracy. When trained on publicly available re-analysis data for the daily temperature of the North American continent, we observe significant improvements over classical baselines such as climatology and persistence-based forecast techniques. Although our experiments concern specific examples, the proposed approach is general, and our results support the viability of kernel methods (with learned kernels) for interpretable and computationally efficient geophysical forecasting for a large diversity of processes.

58 GEOSCIENCES↗
DOE OSTI2023

Fast tree-based algorithms for DBSCAN for low-dimensional data on GPUs

DBSCAN is a well-known density-based clustering algorithm to discover arbitrary shape clusters. While conceptually simple in serial, the algorithm is challenging to efficiently parallelize on manycore GPU architectures. Common pitfalls, such as asynchronous range query calls, result in high thread execution divergence in many implementations. In this paper, we propose a new framework for GPU-accelerated DBSCAN, and describe two tree-based algorithms within that framework. Both algorithms fuse the search for neighbors with updating cluster information, but differ in their treatment of dense regions of the data. We show that the time taken to compute clusters is at most twice that of determination of the neighbors. We compare the proposed algorithms with existing CPU and GPU implementations, and demonstrate their competitiveness and performance using a fast traversal structure (bounding volume hierarchy) for low dimensional data. We also show that the memory usage can be reduced by processing object neighbors dynamically without storing them.

Prokopenko, Andrey↗
DOE OSTI2024

Accelerating cavity fault prediction using deep learning at Jefferson Laboratory

Abstract Accelerating cavities are an integral part of the continuous electron beam accelerator facility (CEBAF) at Jefferson Laboratory. When any of the over 400 cavities in CEBAF experiences a fault, it disrupts beam delivery to experimental user halls. In this study, we propose the use of a deep learning model to predict slowly developing cavity faults. By utilizing pre-fault signals, we train a long short-term memory-convolutional neural network binary classifier to distinguish between radio-frequency (RF) signals during normal operation and RF signals indicative of impending faults. We optimize the model by adjusting the fault confidence threshold and implementing a multiple consecutive window criterion to identify fault events, ensuring a low false positive rate. Results obtained from analysis of a real dataset collected from the accelerating cavities simulating a deployed scenario demonstrate the model’s ability to identify normal signals with 99.99% accuracy and correctly predict 80% of slowly developing faults. Notably, these achievements were achieved in the context of a highly imbalanced dataset, and fault predictions were made several hundred milliseconds before the onset of the fault. Anticipating faults enables preemptive measures to improve operational efficiency by preventing or mitigating their occurrence.

43 PARTICLE ACCELERATORS↗
DOE OSTI2020

PANTHER: A Programmable Architecture for Neural Network Training Harnessing Energy-Efficient ReRAM

The wide adoption of deep neural networks has been accompanied by ever-increasing energy and performance demands due to the expensive nature of training them. Additionally, numerous special-purpose architectures have been proposed to accelerate training: both digital and hybrid digital-analog using resistive RAM (ReRAM) crossbars. ReRAM-based accelerators have demonstrated the effectiveness of ReRAM crossbars at performing matrix-vector multiplication operations that are prevalent in training. However, they still suffer from inefficiency due to the use of serial reads and writes for performing the weight gradient and update step. A few works have demonstrated the possibility of performing outer products in crossbars, which can be used to realize the weight gradient and update step without the use of serial reads and writes. However, these works have been limited to low precision operations which are not sufficient for typical training workloads. Moreover, they have been confined to a limited set of training algorithms for fully-connected layers only. To address these limitations, we propose a bit-slicing technique for enhancing the precision of ReRAM-based outer products, which is substantially different from bit-slicing for matrix-vector multiplication only. We incorporate this technique into a crossbar architecture with three variants catered to different training algorithms. To evaluate our design on different types of layers in neural networks (fully-connected, convolutional, etc.) and training algorithms, we develop PANTHER, an ISA-programmable training accelerator with compiler support. Our design can also be integrated into other accelerators in the literature to enhance their efficiency. Our evaluation shows that PANTHER achieves up to 8.02×, 54.21×, and 103× energy reductions as well as 7.16×, 4.02×, and 16× execution time reductions compared to digital accelerators, ReRAM-based accelerators, and GPUs, respectively.

42 ENGINEERING↗
DOE OSTI2024

Reinforcement Learning as a Parsimonious Alternative to Prediction Cascades: A Case Study on Image Segmentation

Deep learning architectures have achieved state-of-the-art (SOTA) performance on computer vision tasks such as object detection and image segmentation. This may be attributed to the use of over-parameterized, monolithic deep learning architectures executed on large datasets. Although such large architectures lead to increased accuracy, this is usually accompanied by a larger increase in computation and memory requirements during inference. While this is a non-issue in traditional machine learning (ML) pipelines, the recent confluence of machine learning and fields like the Internet of Things (IoT) has rendered such large architectures infeasible for execution in low-resource settings. For some datasets, large monolithic pipelines may be overkill for simpler inputs. To address this problem, previous efforts have proposed decision cascades where inputs are passed through models of increasing complexity until the desired performance is achieved. However, we argue that cascaded prediction leads to sub-optimal throughput and increased computational cost due to wasteful intermediate computations. To address this, we propose PaSeR (Parsimonious Segmentation with Reinforcement Learning) a non-cascading, cost-aware learning pipeline as an efficient alternative to cascaded decision architectures. Through experimental evaluation on both real-world and standard datasets, we demonstrate that PaSeR achieves better accuracy while minimizing computational cost relative to cascaded models. Further, we introduce a new metric IoU/GigaFlop to evaluate the balance between cost and performance. On the real-world task of battery material phase segmentation, PaSeR yields 179% improvement over SOTA MatPhase model and a 196% improvement over IDK Cascades under the IoU/GigaFlop metric. We also demonstrate PaSeR’s adaptability to complementary models trained on a noisy MNIST dataset, where it outperforms all baselines on IoU/GigaFlop by an average of 44%.

Srikshan, Bharat↗
DOE OSTI2024

Batched sparse direct solver design and evaluation in SuperLU_DIST

Over the course of interactions with various application teams, the need for batched sparse linear algebra functions has emerged in order to make more efficient use of the GPUs for many small and sparse linear algebra problems. In this paper, we present our recent work on a batched sparse direct solver for GPUs. The sparse LU factorization is computed by the levels of the elimination tree, leveraging the batched dense operations at each level and a new batched Scatter GPU kernel. The sparse triangular solve is computed by the level sets of the directed acyclic graph (DAG) of the triangular matrix. Batched operations overcome the large overhead associated with launching many small kernels. For medium sized matrix batches with not-so-small bandwidth, using an NVIDIA A100 GPU, our new batched sparse direct solver is orders of magnitude faster than a batched banded solver and uses less than one-tenth of the memory.

Boukaram, Wajih↗