Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “compute in memory”

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

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

At least 271 records · Page 15
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Improved Evaluation of Large Network Matrices for Linear Power Flow Within Optimization Problems: Preprint

This work discusses methods for evaluating the Power Transfer Distribution Factor (PTDF) and Line Outage Distribution Factor (LODF) matrices by employing sparse linear algebra for large-scale computing applications. These matrices are critical in many power systems applications, such as the Unit Commitment Problem (UC), pre- and post-contingency power flow analysis, and transmission expansion. These matrices are typically dense, which means they require a significant amount of time and memory to be computed for large networks. However, by analyzing the structure of the matrices and their computation method, it is possible to use reduced memory methods based on sparse matrix operations. This paper shows that sparse linear algebra algorithms are faster and require less memory and time than traditional dense approaches. Additionally, we explore the effect of matrix sparsification by eliminating trailing digits on power flow calculations.

ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2024

Improved Evaluation of Large Network Matrices for Linear Power Flow Within Optimization Problems

This work presents methods for evaluating the Power Transfer Distribution Factor (PTDF) and Line Outage Distribution Factor (LODF) matrices by employing sparse linear algebra for large-scale computing applications. These matrices play a critical role in many power system applications, such as the Unit Commitment Problem (UC), pre- and post-contingency power flow analysis, and transmission expansion. These matrices are typically dense, which means they require a significant amount of time and memory to be computed for large networks. However, by analyzing the structure of the matrices and their computation method, it is possible to use reduced memory methods based on sparse matrix operations. This paper shows that sparse linear algebra algorithms are faster and require less memory and time than traditional dense approaches. Additionally, we explore the effect of matrix sparsification by eliminating trailing digits on power flow calculations.

large scale↗
DOE OSTI2022

Model reduction methods for nuclear emulators

The field of model order reduction (MOR) is growing in importance due to its ability to extract the key insights from complex simulations while discarding computationally burdensome and superfluous information. We provide an overview of MOR methods for the creation of fast & accurate emulators of memory- and compute-intensive nuclear systems, focusing on eigen-emulators and variational emulators. As an example, we describe how 'eigenvector continuation' is a special case of a much more general and well-studied MOR formalism for parameterized systems. We continue with an introduction to the Ritz and Galerkin projection methods that underpin many such emulators, while pointing to the relevant MOR theory and its successful applications along the way. Here, we believe that this guide will open the door to broader applications in nuclear physics and facilitate communication with practitioners in other fields.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Full-field imaging learning machine (FILM)

A method of determining dynamic properties of a structure (linear or nonlinear) includes receiving spatio-temporal inputs, generating mode shapes and modal components corresponding to the spatio-temporal inputs using a trained deep complexity coding artificial neural network, and subsequently generating the dynamic properties by analyzing each modal component using a trained learning machine. A computing system for non-contact determination of dynamic properties of a structure includes a camera, a processor, and a memory including computer-executable instructions. When the instructions are executed, the system is caused to receive spatio-temporal image data, decompose the spatio-temporal image data into constituent manifold components using an autoencoder, and analyze the constituent manifold components using a trained learning machine to determine the dynamic properties.

Yang, Yongchao↗
DOE OSTI2022

Dynamic voltage frequency scaling based on active memory barriers

A processing unit includes compute units partitioned into one or islands that are provided with operating voltages and clock signals having clock frequencies independent of providing operating voltages or clock signals to other islands of compute units. The processing unit also includes dynamic voltage and frequency scaling (DVFS) hardware configured to compute one or more numbers of active memory barriers in the one or more islands. The DVFS hardware is also configured to modify the operating voltages or clock frequencies provided to the one or more islands in response to a change in numbers of active memory barriers in the one or more islands. In some cases, the operating voltage or clock frequency provided to an island is increased in response to the number of active memory barriers in the island decreasing. The operating voltage or clock frequency provided to the island is decreased in response to the number of active memory barriers in the island increasing.

Bharadwaj, Vedula Venkata Srikant↗
DOE OSTI2022

Towards scaling community detection on distributed-memory heterogeneous systems

Distributed multi-GPU systems pose significant challenges and opportunities for efficient execution of parallel applications. Graph algorithms are generally characterized by irregular memory accesses, low computation to communication ratios, and load balancing problems that are especially hard to address on multi-GPU systems. Graph community detection is an important problem in the emerging domain of graph analytics with numerous applications. In this paper, we present our ongoing work on distributed-memory multi-GPU implementation for graph community detection. Our work parallelizes the widely used (albeit serial) Louvain method on distributed multi-GPU platforms. Supported by an extensive set of experiments on a multi-GPU enabled supercomputer (OLCF Summit) and a single compute node (Nvidia DGX-2®), we demonstrate competitive performance to existing distributed-memory CPU-based implementation, and up to 6.5 better results than Nvidia RAPIDS® CUGRAPH. To the best of our knowledge, this work represents the first effort for community detection on distributed multi-GPU systems. Our approach and related findings can be extended to numerous other iterative graph algorithms on multi-GPU systems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

XploreNAS : Explore Adversarially Robust and Hardware-efficient Neural Architectures for Non-ideal Xbars

Compute In-Memory platforms such as memristive crossbars are gaining focus as they facilitate acceleration of Deep Neural Networks (DNNs) with high area and compute efficiencies. However, the intrinsic non-idealities associated with the analog nature of computing in crossbars limits the performance of the deployed DNNs. Furthermore, DNNs are shown to be vulnerable to adversarial attacks leading to severe security threats in their large-scale deployment. Thus, finding adversarially robust DNN architectures for non-ideal crossbars is critical to the safe and secure deployment of DNNs on the edge. This work proposes a two-phase algorithm-hardware co-optimization approach called XploreNAS that searches for hardware efficient and adversarially robust neural architectures for non-ideal crossbar platforms. We use the one-shot Neural Architecture Search approach to train a large Supernet with crossbar-awareness and sample adversarially robust Subnets therefrom, maintaining competitive hardware efficiency. Our experiments on crossbars with benchmark datasets (SVHN, CIFAR10, CIFAR100) show up to ~8–16% improvement in the adversarial robustness of the searched Subnets against a baseline ResNet-18 model subjected to crossbar-aware adversarial training. We benchmark our robust Subnets for Energy-Delay-Area-Products (EDAPs) using the Neurosim tool and find that with additional hardware efficiency–driven optimizations, the Subnets attain ~1.5–1.6× lower EDAPs than ResNet-18 baseline.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Capturing Time-Varying Wake Dynamics Using Hybridized Actuator Disks in Steady-State Simulations for Improved Optimization Efficiency

When optimizing turbine blade properties, using a time varying, unsteady simulation allows for high-fidelity representations of the wake dynamics. Unfortunately these types of optimizations are prohibitively expensive in terms of computation time and memory requirements. This presentation aims to show a method of transferring the wake dynamics of an unsteady simulation to a much more computationally tractable steady state simulation using hybrid actuator disks informed by unsteady, actuator line model dynamics.

actuator disk model↗
DOE OSTI2026

Scaling kinetic Monte-Carlo simulations of grain growth with combined convolutional and graph neural networks

Graph neural networks (GNN) have emerged as a promising machine learning method for microstructure simulations such as grain growth. However, accurate modeling of realistic grain boundary networks requires large simulation cells, which GNN has difficulty scaling up to. To alleviate the computational costs and memory footprint of GNN, we suggest a hybrid architecture combining a convolutional neural network (CNN) based bijective autoencoder to compress the spatial dimensions, and a GNN that evolves the microstructure in the latent space of reduced spatial sizes. Our results demonstrate that the new design significantly reduces computational costs with using fewer message passing layer (from 12 down to 3) compared with GNN alone. The reduction in computational cost becomes more pronounced as the spatial size increases, indicating strong computational scalability. For the largest mesh evaluated (160 3 ), our method reduces memory usage and runtime in inference by 117× and 115×, respectively, compared with GNN-only baseline. More importantly, it shows higher accuracy and stronger spatiotemporal capability than the GNN-only baseline, especially in long-term testing. Such combination of scalability and accuracy is essential for simulating realistic material microstructures over extended time scales. The improvements can be attributed to the bijective autoencoder’s ability to compress information losslessly from spatial domain into a high dimensional feature space, thereby producing more expressive latent features for the GNN to learn from, while also contributing its own spatiotemporal modeling capability. Training data are generated from stochastic grain growth simulations, providing realistic variability for learning robust microstructure evolution. Comprehensive system validation confirms that the model is accurate, robust, and scalable.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Graph Neural Network-based Tracking as a Service

Recent studies have shown promising results for track finding in dense environments using Graph Neural Network (GNN)-based algorithms. However, GNN-based track finding is computationally slow on CPUs, necessitating the use of coprocessors to accelerate the inference time. Additionally, the large input graph size demands a large device memory for efficient computation, a requirement not met by all computing facilities used for particle physics experiments, particularly those lacking advanced GPUs. Furthermore, deploying the GNN-based track-finding algorithm in a production environment requires the installation of all dependent software packages, exclusively utilized by this algorithm. These computing challenges must be addressed for the successful implementation of GNN-based track-finding algorithm into production settings. In response, we introduce a ``GNN-based tracking as a service'' approach, incorporating a custom backend within the NVIDIA Triton inference server to facilitate GNN-based tracking. This paper presents the performance of this approach using the Perlmutter supercomputer at NERSC.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

COLLABORATIVE DEVELOPMENT PROJECTS - PHOTONIC MEMORY CONTROLLER MODULE (P-MCM)

As computational density for high-performance computing and big-data services continues to scale, performance scalability of next generation computing systems is becoming increasingly constrained by limitations in memory access, power dissipation and chip packaging. The processor-memory communication bottleneck, a major challenge in current multicore processors due to limited pin-out and power budget, presents a detrimental scaling barrier to data-intensive computing. A consortium team of small businesses and leading researchers that includes experts from photonics processor-memory architecture, III/V photonic laser design/fabrication, silicon photonics design/fabrication, photonics packaging and assembly, and FPGA-based high-performance memory controller IP development – to collaboratively develop a commercialization path for a Photonic Memory Controller Module (P-MCM).

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Performance Study of CXL Memory Topology

This paper presents a comprehensive evaluation of the performance impact of various Compute Express Link (CXL) memory topologies, with a particular emphasis on CXL switches, in the context of High- Performance Computing (HPC) and Large Language Model (LLM) inference workloads. Our study unveils significant performance variations across different topologies, demonstrating that certain configurations yield superior performance for specific workloads. These findings underscore the critical importance of tailored topol- ogy selection in optimizing system performance. Additionally, we address the inherent challenges associated with integrating CXL switches, including overhead considerations and routing complex- ities. Our research highlights the necessity for thorough evalua- tion methodologies to fully leverage CXL technology’s potential in contemporary computing environments. These insights provide valuable guidance for system architects and data center operators in designing and optimizing CXL-based infrastructures for diverse workload requirements.

CXL, memory, Artificial Intelligence (AI), HPC↗
DOE OSTI2025

Moment-based adaptive time integration for thermal radiation transport

Here, in this paper we develop a framework for moment-based adaptive time integration of deterministic multifrequency thermal radiation transpot (TRT). We generalize our recent semi-implicit-explicit (IMEX) integration framework for gray TRT to multifrequency TRT, and also introduce a semi-implicit variation that facilitates higher-order integration of TRT, where each stage is implicit in all components except opacities. To appeal to the broad literature on adaptivity with Runge–Kutta methods, we derive new embedded methods for four asymptotic preserving IMEX Runge–Kutta schemes we have found to be robust in our previous work on TRT and radiation hydrodynamics. We then use a moment-based high-order-low-order representation of the transport equations. Due to the high dimensionality, memory is always a concern in simulating TRT. We form error estimates and adaptivity in time purely based on temperature and radiation energy, for a trivial overhead in computational cost and memory usage compared with the base second order integrators. We then test the adaptivity in time on the tophat and Larsen problem, demonstrating the ability of the adaptive algorithm to naturally vary the timestep across 4–5 orders of magnitude, ranging from the dynamical timescales of the streaming regime to the thick diffusion limit.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

MICCO: An Enhanced Multi-GPU Scheduling Framework for Many-Body Correlation Functions

Calculation of many-body correlation functions is one of the critical kernels utilized in many scientific computing areas, especially in Lattice Quantum Chromodynamics (Lattice QCD). It is formalized as a sum of a large number of contraction terms each of which can be represented by a graph consisting of vertices describing quarks inside a hadron node and edges designating quark propagations at specific time intervals. Due to its computation- and memory-intensive nature, real-world physics systems (e.g., multi-meson or multi-baryon systems) explored by Lattice QCD prefer to leverage multi-GPUs. Different from general graph processing, many-body correlation function calculations show two specific features: a large number of computation-/data-intensive kernels and frequently repeated appearances of original and intermediate data. The former results in expensive memory operations such as tensor movements and evictions. The latter offers data reuse opportunities to mitigate the data-intensive nature of many-body correlation function calculations. However, existing graph-based multi-GPU schedulers cannot capture these data-centric features, thus resulting in a sub-optimal performance for many-body correlation function calculations. To address this issue, this paper presents a multi-GPU scheduling framework, MICCO, to accelerate contractions for correlation functions particularly by taking the data dimension (e.g., data reuse and data eviction) into account. This work first performs a comprehensive study on the interplay of data reuse and load balance, and designs two new concepts: local reuse pattern and reuse bound to study the opportunity of achieving the optimal trade-off between them. Based on this study, MICCO proposes a heuristic scheduling algorithm and a machine-learning-based regression model to generate the optimal setting of reuse bounds. Specifically, MICCO is integrated into a real-world Lattice QCD system, Redstar, for the first time running on multiple GPUs. The evaluation demonstrates MICCO outperforms other state-of-art works, achieving up to 2.25× speedup in synthesized datasets, and 1.49× speedup in real-world correlation functions.

Wang, Qihan↗
DOE OSTI2024

Performance Improvements of the Griffin Solvers in FY24

The Griffin code is a MOOSE-based reactor physics application jointly developed by Idaho National Laboratory and Argonne National Laboratory under the Department of Energy Office of Nuclear Energy Nuclear Energy Advanced Modeling and Simulation Program. This fiscal year, we have made significant efforts to improve the performance of transport solver options and cross-section generation for the efficient use of Griffin in advanced reactor applications. For the HFEM-PN solver, the residual evaluations of HFEM kernels were optimized by utilizing the pre- computed averaged cross sections for individual elements. Numerical integration involving the evaluation of basis functions at quadrature points was bypassed by facilitating precomputed element mass matrices for response matrices. Red-black iterations were improved by introducing a new generalized minimum residual based solver. The memory usage of response matrix storage was significantly reduced by applying basis function rotations on interfaces and calculating volumetric odd-parity moments on the fly. Additionally, the adjoint flux and transient calculation capabilities of the HFEM-PN solver were successfully implemented and verified using the TWIGL benchmark problem. For the DFEM-SN solver, memory footprint and computation time were significantly reduced by not treating angular flux vectors as the MOOSE nonlinear system vectors. Specifically for IQS, scalar adjoint weighting was introduced to further eliminate angular adjoint flux storage in the MOOSE auxiliary system. It was demonstrated through the three-dimensional Advanced Burner Test Reactor core problem that the memory usage for transient calculations with the IQS method was reduced by over 7.5× compared to before the optimizations. For the self-shielding application programming interface, a new double-heterogeneity treatment method, named the Bell Function-Based Analytic Two-Region Slowing Down Method, was developed to efficiently flux-volume homogenize TRISO particles with the matrix. Additionally, optimizations were made to hyper- fine group (HFG) slowing down calculations by pretabulating collision probability coefficients and grouping isotopes, significantly reducing the computational time for calculating scattering sources per HFG. Lastly, the pin power reconstruction module was extended to account for temporal behavior in a microreactor analysis problem, specifically for a control drum transient. Verification tests for each of these improvements demonstrated significant performance enhancements and memory reduction.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2021

Toward the Minimal Floating Operation Count Cholesky Decomposition of Electron Repulsion Integrals

As quantum chemistry calculations deal with molecular systems of increasing size, the memory requirement to store electron-repulsion integrals (ERIs) greatly outpaces the physical memory available in computing hardware. Here, the Cholesky decomposition of ERIs provides a convenient yet accurate technique to reduce the storage requirement of integrals. Recent developments of a two-step algorithm have drastically reduced the memory operation (MOP) count, leaving the floating operation (FLOP) count as the last frontier of cost reduction in the Cholesky ERI algorithm. In this report, we introduce a dynamic integral tracking, reusing, and compression/elimination protocol embedded in the two-step Cholesky ERI method. Benchmark studies suggest that this technique becomes particularly advantageous when the basis set consists of many computationally expensive high-angular-momentum basis functions. With this dynamic ERI improvement, the Cholesky ERI approach proves to be a highly efficient algorithm with minimal FLOP and MOP count.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2022

Early Performance Results on 4th Gen Intel(R) Xeon (R) Scalable Processors with DDR and Intel(R) Xeon(R) processors, codenamed Sapphire Rapids with HBM

The Crossroads supercomputer was designed to simulate some of the most complex physical devices in the world. These simulations routinely require 1/2 petabyte or more of system memory running on thousands of compute nodes for months at a time on the most powerful supercomputers. Improvements in time to solutions for these workloads can have major impact on our mission capabilities. In this paper we present early results of representative application workloads on 4th Gen Intel Xeon and Intel Xeon Processors codenamed Sapphire Rapids with HBM. These results demonstrate an extremely promising 8.57x improvement (node to node) over our prior generation Intel Broadwell (BDW) based HPC systems. No code modifications were required to achieve this speedup, providing a compelling path forward toward major reductions in time to solution and the complexity of physical systems that can be simulated in the future.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

A comparative study of two numerical approaches for solving Kim–Kim–Suzuki phase-field models

Among the standard multi-phase multi-component phase-field (PF) methods, the Kim–Kim–Suzuki (KKS) method has the advantage of decoupling interfacial energy from bulk energy and solving concentration as the conserved variable. There are two approaches to numerically solving a KKS method: the global solution approach (GSA) solves all variables in a global system simultaneously, and the local solution approach (LSA) solves phase concentrations locally using a Newton solver. This work compares the performance of LSA and GSA for solving four KKS models of increasing complexity with the finite element method using the MOOSE framework. The solution accuracy, degrees of freedom (DOFs), memory usage, and computational efficiency are compared. We find that GSA and LSA generate similar solutions, with a maximum difference of only 0.34%. For each model, LSA has a lower number of DOFs, utilizes less memory, and less wall time. Additionally, the savings of memory and wall time in LSA increase with increasing mesh density of the same model and are more pronounced in models with higher dimensionality and more nodes. However, GSA is easier to implement in existing codes and can better solve highly nonlinear systems by utilizing sophisticated solvers.

36 MATERIALS SCIENCE↗