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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Approaches for the Simulation of Coupled Processes in Evolving Fractured Porous Media Enabled by Exascale Computing

Models have historically represented fractured porous media with continuum descriptions that characterize the media using bulk parameters. The impact of small-scale features is not captured in these models, although they may be controlling the performance of subsurface applications. Pore-scale models can simulate processes in small-scale features by representing the pore space geometry explicitly but are computationally expensive for large domains. The alternative multiscale approach entails the combination of pore-scale and continuum-scale descriptions in a single framework. We use Chombo-Crunch, a computational capability that discretizes complex geometries with an adaptive, embedded boundary method to contrast these two approaches. Chombo-Crunch takes advantage of recent computational performance and memory bandwidth improvements resulting from the emergence of exascale computing resources. These combined improvements enable the efficient simulation of reactive transport in fractured media with a high degree of fidelity and the ability to capture the control small-scale processes exert on the overall medium evolution.

42 ENGINEERING↗
DOE OSTI2024

Hutchinson Trace Estimation for high-dimensional and high-order Physics-Informed Neural Networks

Physics-Informed Neural Networks (PINNs) have proven effective in solving partial differential equations (PDEs), especially when some data are available by seamlessly blending data and physics. However, extending PINNs to high-dimensional and even high-order PDEs encounters significant challenges due to the computational cost associated with automatic differentiation in the residual loss function calculation. Herein, we address the limitations of PINNs in handling high-dimensional and high-order PDEs by introducing the Hutchinson Trace Estimation (HTE) method. Starting with the second-order high-dimensional PDEs, which are ubiquitous in scientific computing, HTE is applied to transform the calculation of the entire Hessian matrix into a Hessian vector product (HVP). This approach not only alleviates the computational bottleneck via Taylor-mode automatic differentiation but also significantly reduces memory consumption from the Hessian matrix to an HVP’s scalar output. We further showcase HTE’s convergence to the original PINN loss and its unbiased behavior under specific conditions. Comparisons with the Stochastic Dimension Gradient Descent (SDGD) highlight the distinct advantages of HTE, particularly in scenarios with significant variability and variance among dimensions. We further extend the application of HTE to higher-order and higher-dimensional PDEs, specifically addressing the biharmonic equation. By employing tensor-vector products (TVP), HTE efficiently computes the colossal tensor associated with the fourth-order high-dimensional biharmonic equation, saving memory and enabling rapid computation. The effectiveness of HTE is illustrated through experimental setups, demonstrating comparable convergence rates with SDGD under memory and speed constraints. Additionally, HTE proves valuable in accelerating the Gradient-Enhanced PINN (gPINN) version as well as the Biharmonic equation. Overall, HTE opens up a new capability in scientific machine learning for tackling high-order and high-dimensional PDEs.

Curse of dimensionality↗
DOE OSTI2021

GPU-based Image Compression for Efficient Compositing in Distributed Rendering Applications

Visualizations of large-scale data sets are often created on graphics clusters that distribute the rendering task amongst many processes. When using real-time GPU-based graphics algorithms, the most time-consuming aspect of distributed rendering is typically the com-positing phase - combining all partial images from each rendering process into the final visualization. Compo siting requires image data to be copied off the GPU and sent over a network to other processes. While compression has been utilized in existing distributed rendering compositors to reduce the data being sent over the network, this compression tends to occur after the raw images are transferred from the GPU to main memory. In this paper, we present work that leverages OpenGL / CUDA interoperability to compress raw images on the GPU prior to transferring the data to main memory. This approach can significantly reduce the device-to-host data transfer time, thus enabling more efficient compositing of images generated by distributed rendering applications.

Lipinksi, Riley↗
DOE OSTI2024

Performance Evaluation of District Energy Microgrids Planning Tool for Non-Technical Users

Community Microgrids are increasingly gaining popularity worldwide for their efficiency, cost-effectiveness, and local resilience improvement. Microgrid planning tools play a crucial role in their deployment. In the process, tentative designs of the microgrid are simulated, analyzed, and optimized. Due to the complexity of the problem, planning tools must carefully balance computational efficiency while seeking the most optimal solutions. This paper investigates the impact of algorithm selection on CPU and memory utilization of a Community Microgrid planning tool that is specifically designed for non-technical users. We compare two version of the code with two alternatives for Community Microgrid planning tools: a Greedy Algorithm, and a Linear Programming Approach. This comparison examines scenarios spanning from 2 to 50 buildings. Our findings revealed a significant difference in performance between the two algorithms, underscoring the critical role of algorithm selection in optimizing the efficiency of Community Microgrid Planning Tools.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2026

Design, characterization and shape recovery behavior of 3D/4D printed shape memory polymers (SMPs)

Shape memory polymers (SMPs) represent a paradigm shift in material science, uniquely capable of undergoing reversible shape transformations triggered by external stimuli, positioning them as pivotal in developing next-generation biomedical devices, aerospace components, and adaptive structures. Extensive research has been done on SMPs with a major focus on high-temperature programming methods, which can limit energy efficiency and applicability with temperature-sensitive materials. Additionally, while various SMP blends have demonstrated great potential, limited work has been done on the suitability for 3D printing these materials, particularly under high-strain and ambient temperature programming conditions. In this study, a three-component optimized SMP composition was evaluated by 3D printing via the Material Extrusion (MEX) technique and investigating its ambient temperature-programming behavior at high strains. The SMP formulation studied was a tailored blend of thermoplastic polyurethane (TPU), polycaprolactone (PCL), and an octadecane diol-based copolymer (OBC) that exhibits robust shape memory behavior, high strain tolerance, and efficient force generation. Rigorous thermal, mechanical, and shape recovery analyses, along with optimized printing parameters and consistent shape recovery rates of up to 90%, were achieved under dynamic mechanical analysis (DMA), even under ambient programming conditions. This work demonstrates the SMP composition’s potential for adaptive, self-deployable systems with 4D printing characteristics ideal for bio-inspired structures and artificial muscle fibers.

Sudan, Kavish [University of Louisville, KY]↗
DOE OSTI2025

Fast Sideband Control of a Weakly Coupled Multimode Bosonic Memory

Circuit quantum electrodynamics (cQED) with superconducting cavities coupled to nonlinear circuits like transmons offers a promising platform for hardware-efficient quantum information processing. We address critical challenges in realizing this architecture by weakening the dispersive coupling while also demonstrating fast, high-fidelity multimode control by dynamically amplifying gate speeds through transmon-mediated sideband interactions. This approach enables transmon-cavity SWAP gates, for which we achieve speeds up to 30 times larger than the bare dispersive coupling. Combined with transmon rotations, this allows for efficient, universal state preparation in a single cavity mode, though achieving unitary gates and extending control to multiple modes remains a challenge. In this work, we overcome this by introducing two sideband control strategies: (1) a shelving technique that prevents unwanted transitions by temporarily storing populations in sideband-transparent transmon states and (2) a method that exploits the dispersive shift to synchronize sideband transition rates across chosen photon-number pairs to implement transmon-cavity SWAP gates that are selective on photon number. We leverage these protocols to prepare Fock and binomial code states across any of ten modes of a multimode cavity with millisecond cavity coherence times. We demonstrate the encoding of a qubit from a transmon into arbitrary vacuum and Fock state superpositions, as well as entangled NOON states of cavity mode pairs— a scheme extendable to arbitrary multimode Fock encodings. Furthermore, we implement a new binomial encoding gate that converts arbitrary transmon superpositions into binomial code states in $\qty{4}{\micro\second}$ (less than $1/\chi$), achieving an average post-selected final state fidelity of $\qty{96.3}{\percent}$ across different fiducial input states.

Huang, Jordan [Rutgers U., Piscataway]↗
DOE OSTI2020

Pass-efficient methods for compression of high-dimensional turbulent flow data

The future of high-performance computing, specifically on future Exascale computers, will presumably see memory capacity and bandwidth fail to keep pace with data generated, for instance, from massively parallel partial differential equation (PDE) systems. Current strategies proposed to address this bottleneck entail the omission of large fractions of data, as well as the incorporation of in situ compression algorithms to avoid overuse of memory. To ensure that post-processing operations are successful, this must be done in a way that a sufficiently accurate representation of the solution is stored. Moreover, in situations where the input/output system becomes a bottleneck in analysis, visualization, etc., or the execution of the PDE solver is expensive, the number of passes made over the data must be minimized. In the interest of addressing this problem, this work focuses on the utility of pass-efficient, parallelizable, low-rank, matrix decomposition methods in compressing high-dimensional simulation data from turbulent flows. Additionally, a particular emphasis is placed on using coarse representation of the data – compatible with the PDE discretization grid – to accelerate the construction of the low-rank factorization. This includes the presentation of a novel single-pass matrix decomposition algorithm for computing the so-called interpolative decomposition. The methods are described extensively and numerical experiments on two turbulent channel flow data are performed. In the first (unladen) channel flow case, compression factors exceeding 400 are achieved while maintaining accuracy with respect to first- and second-order flow statistics. In the particle-laden case, compression factors of 100 are achieved and the compressed data is used to recover particle velocities. These results show that these compression methods can enable efficient computation of various quantities of interest in both the carrier and disperse phases.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

A Low-Rank QTT-based Finite Element Method for Elasticity Problems

We present an efficient and robust numerical algorithm for solving the linear elasticity problem that combines the Quantized Tensor Train format and a domain partitioning strategy. This approach makes it possible to solve the linear elasticity problem on a computational domain that is more general than a square. By integrating Z-ordering and subdomain concatenation, our method substantially decreases memory usage and achieves a notable reduction in rank compared to established Finite Element implementations like the FEniCS platform. This efficiency is maintained while still guaranteeing exponential convergence with respect to the number of degrees of freedom. This performance gain, however, requires a fundamental rethinking of how core finite element operations are implemented. This includes changes to mesh discretization, node and degree of freedom ordering, stiffness matrix and internal nodal force assembly, and the execution of algebraic matrix-vector operations. In this work, we discuss all these aspects in detail and assess the method’s performance in the numerical approximation of three representative test cases.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Improving deep learning performance for predicting large-scale geological ${{CO}_{2}}$ sequestration modeling through feature coarsening

Physics-based reservoir simulation for fluid flow in porous media is a numerical simulation method to predict the temporal-spatial patterns of state variables (e.g. pressure p) in porous media, and usually requires prohibitively high computational expense due to its non-linearity and the large number of degrees of freedom (DoF). This work describes a deep learning (DL) workflow to predict the pressure evolution as fluid flows in large-scale 3-dimensional(3D) heterogeneous porous media. In particular, we develop an efficient feature coarsening technique to extract the most representative information and perform the training and prediction of DL at the coarse scale, and further recover the resolution at the fine scale by spatial interpolation. We validate the DL approach to predict pressure field against physics-based simulation data for a field-scale 3D geologic CO 2 sequestration reservoir model. We evaluate the impact of feature coarsening on DL performance, and observe that the feature coarsening not only decreases the training time by >74% and reduces the memory consumption by >75%, but also maintains temporal error 0.63% on average. Besides, the DL workflow provides predictive efficiency with 1406 times speedup compared to physics-based numerical simulation. The key findings from this research significantly improve the training and prediction efficiency of deep learning model to deal with large-scale heterogeneous reservoir models, and thus it can also be further applied to accelerate workflows of history matching and reservoir optimization for close-loop reservoir management.

58 GEOSCIENCES↗
DOE OSTI2026

Ensemble Kalman filter for data assimilation coupled with low-resolution computations techniques applied in fluid dynamics

This paper presents an innovative Reduced-order model (ROM) for merging experimental and simulation data using data assimilation (DA) to estimate the "True" state of a fluid dynamics system, leading to more accurate predictions. Our methodology introduces a novel approach by implementing the ensemble Kalman filter (EnKF) within a reduced-dimensional framework, grounded in a robust theoretical foundation and applied to fluid dynamics. To address the substantial computational demands of DA, the proposed ROM employs low-resolution (LR) techniques to drastically reduce computational costs. This innovative approach involves downsampling datasets for DA computations, followed by an advanced reconstruction technique based on low-cost singular value decomposition (lcSVD). The lcSVD method, a key innovation in this paper, has never been applied to DA before and offers a highly efficient way to enhance resolution with minimal computational resources. Our results demonstrate significant reductions in both computation time and RAM usage through these LR techniques without compromising the accuracy of the estimations. For instance, in a turbulent test case, for a data compression rate of 15.9, the LR approach can achieve a speed-up of 13.7 and a RAM compression of 90.9% while maintaining a low relative root mean square error (RRMSE) of 2.6%, compared to 0.8% in the high-resolution (HR) reference. Furthermore, we highlight the effectiveness of the EnKF in estimating and predicting the state of fluid flow systems based on limited observations and given low-fidelity numerical data. This paper highlights the potential of the proposed DA method in fluid dynamics applications, particularly for improving computational efficiency in CFD and related fields. Its ability to balance accuracy with low computational and memory costs makes it especially suitable for large-scale and real-time applications, such as environmental monitoring or engineering design. This method will be incorporated into ModelFLOWs-app.

Data Assimilation↗
DOE OSTI2023

Quantitative proton radiography and shadowgraphy for arbitrary intensities

Charged-particle radiography and shadowgraphy data can be directly inverted to obtain a line-integrated transverse Lorentz force or a line-integrated transverse refractive index gradient if intensity modulations due to scattering and absorption are negligible, and angular deflections are small. We develop a new direct-inversion algorithm based on plasma physics and compare it to a new Monge–Ampère code and an existing power diagram code. The measured or source intensity is represented by electrons subject to drag, and the other intensity by fixed ions. The decrease in kinetic plus electrostatic energy determines convergence. The displacement of the electrons from their initial to their equilibrium positions determines the line-integrated force or refractive index gradient. We have implemented two approaches: PIC (particle in cell) and Lagrangian fluid, in 1-D and 2-D. The PIC code works for arbitrary intensities, can work efficiently in parallel, and can make use of existing codes. The Lagrangian code requires less memory and is faster than the PIC code without massively parallel processing, but fails in 2-D for large intensity modulations. The Monge–Ampère code is by far the fastest in 2-D, without massively parallel processing, but fails for intensities with large voids, high contrast ratios and large deflections across the boundaries, and could not obtain the degree of convergence possible with the PIC code. As a result, the power diagram code was by far the slowest and most memory intensive, and failed for large peaks in the measured intensity.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Convergence study of wakefield simulations with GdfidL and ECHO3D

The interaction of charged particle beams with vacuum chamber components gives rise to electromagnetic wakefields, whose frequency-domain representation is known as beam coupling impedance. Geometric impedance arising from discontinuities and transitions in the vacuum chamber is the focus of this study. Minimizing this impedance is essential to mitigate adverse collective effects in modern storage rings operating with high-intensity particle beams. Accurate and reliable impedance simulations is a key factor of the vacuum chamber design. This paper presents the results of a convergence study of two widely used electromagnetic solvers, GdfidL and ECHO3D, applied to key vacuum-chamber components of the National Synchrotron Light Source II (NSLS-II) storage ring. Detailed comparisons are performed for several geometries, including flange absorbers, RF bellows, button-type beam position monitors, and an in-vacuum undulator (IVU). The results show notable differences in convergence and computational efficiency between the two codes. While GdfidL provides highly resolved results and serves as a common benchmark tool, ECHO3D yields consistent results with coarser meshes, significantly reducing simulation time and memory demands. Simulations with a full-geometry IVU model demonstrate that simplified taper-transition models can miss important impedance contributions. In conclusion, these findings provide practical guidelines for efficient and accurate impedance modeling to optimize design of vacuum chamber components for accelerators.

36 MATERIALS SCIENCE↗
DOE OSTI2024

The Effective Fragment Molecular Orbital Method: Achieving High Scalability and Accuracy for Large Systems

The effective fragment molecular orbital (EFMO) method has been developed to predict the total energy of a very large molecular system accurately (with respect to the underlying quantum mechanical method) and efficiently by taking advantage of the locality of strong chemical interactions and employing a two-level hierarchical parallelism. The accuracy of the EFMO method is partly attributed to the accurate and robust intermolecular interaction prediction between distant fragments, in particular, the many-body polarization and dispersion effects, which require the generation of static and dynamic polarizability tensors by solving the coupled perturbed Hartree–Fock (CPHF) and time-dependent HF (TDHF) equations, respectively. Solving the CPHF and TDHF equations is the main EFMO computational bottleneck due to the inefficient (serial) and I/O-intensive implementation of the CPHF and TDHF solvers. In this work, the efficiency and scalability of the EFMO method are significantly improved with a new CPU memory-based implementation for solving the CPHF and TDHF equations that are parallelized by either message passing interface (MPI) or hybrid MPI/OpenMP. Here, the accuracy of the EFMO method is demonstrated for both covalently bonded systems and noncovalently bound molecular clusters by systematically examining the effects of basis sets and a key distance-related cutoff parameter, R cut . R cut determines whether a fragment pair (dimer) is treated by the chosen ab initio method or calculated using the effective fragment potential (EFP) method (separated dimers). Decreasing the value of Rcut increases the number of separated (EFP) dimers, thereby decreasing the computational effort. It is demonstrated that excellent accuracy (<1 kcal/mol error per fragment) can be achieved when using a sufficiently large basis set with diffuse functions coupled with a small R cut value. With the new parallel implementation, the total EFMO wall time is substantially reduced, especially with a high number of MPI ranks. Given a sufficient workload, nearly ideal strong scaling is achieved for the CPHF and TDHF parts of the calculation. For the first time, EFMO calculations with the inclusion of long-range polarization and dispersion interactions on a hydrated mesoporous silica nanoparticle with explicit water solvent molecules (more than 15k atoms) are achieved on a massively parallel supercomputer using nearly 1000 physical nodes. In addition, EFMO calculations on the carbinolamine formation step of an amine-catalyzed aldol reaction at the nanoscale with explicit solvent effects are presented.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Ab initio generalized Langevin equation

We introduce a machine learning–based approach called ab initio generalized Langevin equation (AIGLE) to model the dynamics of slow collective variables (CVs) in materials and molecules. In this scheme, the parameters are learned from atomistic simulations based on ab initio quantum mechanical models. Force field, memory kernel, and noise generator are constructed in the context of the Mori–Zwanzig formalism, under the constraint of the fluctuation–dissipation theorem. Combined with deep potential molecular dynamics and electronic density functional theory, this approach opens the way to multiscale modeling in a variety of situations. Here, we demonstrate this capability with a study of two mesoscale processes in crystalline lead titanate, namely the field-driven dynamics of a planar ferroelectric domain wall, and the dynamics of an extensive lattice of coarse-grained electric dipoles. In the first case, AIGLE extends the reach of ab initio simulations to a regime of noise-driven motions not accessible to molecular dynamics. In the second case, AIGLE deals with an extensive set of CVs by adopting a local approximation for the memory kernel and retaining only short-range noise correlations. The scheme is computationally more efficient than molecular dynamics by several orders of magnitude and mimics the microscopic dynamics at low frequencies where it reproduces accurately the dominant far-infrared absorption frequency.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Operator evolution from the similarity renormalization group and the Magnus expansion

The Magnus expansion is an efficient alternative to solving similarity renormalization group (SRG) flow equations with high-order, memory-intensive ordinary differential equation solvers. The numerical simplifications it offers for operator evolution are particularly valuable for in-medium SRG calculations, though challenges remain for difficult problems involving intruder states. Here we test the Magnus approach in an analogous but more accessible situation, which is the free-space SRG treatment of the spurious bound states arising from a leading-order chiral effective field theory (EFT) potential with very high cutoffs. In this work, we show that the Magnus expansion passes these tests and then use the investigations as a springboard to address various aspects of operator evolution that have renewed relevance in the context of the scale and scheme dependence of nuclear processes. These aspects include SRG operator flow with band- versus block-diagonal generators, universality for chiral EFT Hamiltonians and associated operators with different regularization schemes, and the impact of factorization arising from scale separation. Implications for short-range correlations physics and the possibilities for reconciling high- and low-resolution treatments of nuclear structure and reactions are discussed.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

GENESPACE R Package (GENESPACE) v1.0

In short, the GENESPACE pipeline conducts analysis of orthology networks, constrained within syntenic regions. Since analyses are limited to local tests conducted within syntenic blocks, GENESPACE is agnostic to ploidy, duplicated regions, inversions or other whole-genome chromosomal complexities that are common across many evolutionary lineages. This advantage allows for evolutionary tests in polyploids (e.g. switchgrass, manuscript in review), species with ancient, but retained whole-genome duplications (e.g. pecan, manuscript in prep), high levels of tandem array proliferation (e.g. eukalypts, manuscript in review) and many other factors that can confound comparative genomic analyses. The major advances of GENESPACE are three-fold: First, this is the first R package to integrate visualization and analysis of large-scale comparative genomics. R, which offers a high-level environment for graphical and statistical exploration of data, is often speed- and memory-limited and not used for computationally intensive tasks such as comparative genomics. The highly efficient C++ scripts used in GENESPACE (via data.table) permit a much faster and computationally lightweight implementation of comparative genomics than is currently available. Second, the pipeline itself is novel. To the best of our knowledge, no other program accomplishes synteny-constrained and ploidy-agnostic comparative genomics. Since nearly all plants and many animals have a history of whole-genome duplications, this is a major and necessary advance to the field. Third, GENESPACE offers high-level and intuitive multi-genome graphical outputs. The dotplots and 'riparian' plots produced herein, which are produced entirely through original R code, are publication-ready and easily customizable.

Schmutz, Jeremy↗
DOE OSTI2020

Investigation of spin and magnetism in two dimensional materials with atomic-scale spatial resolution (Final Technical Report)

Spintronics is an approach to energy-efficient electronics beyond silicon CMOS that utilizes the electron’s spin degree of freedom for memory and logic. The record spin diffusion length in graphene has generated considerable interest in two dimensional (2D) materials (i.e. graphene and beyond) as a platform for spintronics. Magnetism in 2D van der Waals materials may offer additional opportunities for tunability via strain, external fields, and layer-dependent quantum effects. The inherently atomic-scale nature of exchange interactions and the inherently all-surface nature of 2D sheets motivate an integrated approach combining atomic-scale magnetic characterization, epitaxial growth, and nanodevice fabrication. The mission of the research is to develop and utilize atomic-scale imaging of spin and magnetism via spin-polarized scanning tunneling microscopy (STM) in conjunction with atom-by-atom materials growth via molecular beam epitaxy (MBE) and nanoscale device fabrication to explore new frontiers of spintronics and quantum science in 2D materials. The PIs are particularly motivated by understanding how the magnetic and spin-dependent properties at the atomic scale lead to interesting and potentially useful phenomena in spintronic devices at the macroscopic scale.

36 MATERIALS SCIENCE↗
DOE OSTI2024

AXI4MLIR: User-Driven Automatic Host Code Generation for Custom AXI-Based Accelerators

Tensor algebra operations represent an important class of algorithms used across many applications, including machine learning, scientific computing, and data analytics. As a result, the efficient generation of custom accelerators for tensor operations has received increased attention. Previous efforts have produced automated tools enabling users to prototype and explore optimized accelerators. However, little effort has been focused on the host-accelerator interaction in these tools. Efficient use of hardware accelerators requires knowledge about the accelerator's capabilities (operations, data formats, and opcode support), the host CPU microarchitecture (e.g., memory hierarchy), the host-accelerator interface, and the application's features (which code regions should be mapped onto an accelerator). Manually rewriting the original applications to facilitate improved custom accelerator mapping is an error-prone and time-consuming endeavor. To cope with this, we propose AXI4MLIR, a new framework to automatically generate and optimize the communication between the host CPU and arbitrary accelerators that implement linear algebra algorithms. AXI4MLIR extends the MLIR compiler framework to automatically generate efficient host-accelerator driver code for accelerators with AXI-based interfaces. Our compiler extensions enable automatic driver code generation while carefully considering the host's memory hierarchy and target accelerator features. To demonstrate the flexibility and utility of AXI4MLIR, we test it with diverse use cases that include different types of accelerators, tiling scenarios, and dataflow schemes. We compare our experimental results to manual implementations of host-accelerator driver code and find that our approach can reduce CPU cache references by 56% and deliver up to a 1.65x speedup.

Bohm Agostini, Nicolas↗