Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Software Requirements”

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 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Single-Bar SMT Creep-Fatigue Testing on Alloy 617 with Software Controls on elastic Follow-Up Feedback in Support of New Creep-Fatigue Design Methodology

The Simplified Model Test (SMT) is one of the alternative ways to calculate the creep-fatigue damage of elevated temperature components. The SMT approach unifies the creep and fatigue damage, simplifying the damage calculation process with improved accuracy. Traditionally, SMT tests has have been conducted through the two bars: first is the driver bar and second is the test bar. The driver bar stays elastic and imposes the confinement to the second bar. This experimental procedure requires two test frames and a large test specimen, which limits the test parameter range and accessibility of this testing procedures. Hence, a single-bar SMT (SB-SMT) tests has been developed to simplify complexity in SMT experiment setup. The software-controlled SB-SMT test process has been introduced which completely replaces the driver bar. This paper discusses the single-bar SMT (SB-SMT) test procedure with software controls. The challenges and critical parameters in the software-controlled SB-SMT procedure are addresses addressed and recommendations are provided. A scoping test with a set of wider strain range, elastic follow-ups and dwell time validated the proposed test procedure.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2024

Matilda v1.0: An R package for probabilistic climate projections using a reduced complexity climate model

A primary advantage to using reduced complexity climate models (RCMs) has been their ability to quickly conduct probabilistic climate projections, a key component of uncertainty quantification in many impact studies and multisector systems. Providing frameworks for such analyses has been a target of several RCMs used in studies of the future co-evolution of the human and Earth systems. In this paper, we present Matilda, an open-science R software package that facilitates probabilistic climate projection analysis, implemented here using the Hector simple climate model in a seamless and easily applied framework. The primary goal of Matilda is to provide the user with a turn-key method to build parameter sets from literature-based prior distributions, run Hector iteratively to produce perturbed parameter ensembles (PPEs), weight ensembles for realism against observed historical climate data, and compute probabilistic projections for different climate variables. This workflow gives the user the ability to explore viable parameter space and propagate uncertainty to model ensembles with just a few lines of code. The package provides significant freedom to select different scoring criteria and algorithms to weight ensemble members, as well as the flexibility to implement custom criteria. Additionally, the architecture of the package simplifies the process of building and analyzing PPEs without requiring significant programming expertise, to accommodate diverse use cases. We present a case study that provides illustrative results of a probabilistic analysis of mean global surface temperature as an example of the software application.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Remote operation of the DIII-D National Fusion Facility

Abstract Full remote scientific operation of the DIII-D National Fusion Facility is now possible through significant advances in the computer science hardware and software infrastructure made over the last decade. Capabilities around information visualization, data movement, and communication have all been enhanced. The level of capability deployed to remotely operate DIII-D required an infrastructure advancement over what had previously been achieved in the fusion community. The large quantity of real-time data that is automatically displayed on DIII-D’s control room screens can now be visualized by remote participants via web-based applications. New audio/video solutions using the VoIP and instant messaging application Discord have been implemented to mimic the dynamic and ad-hoc scientific conversations that are critical in successfully operating an experimental campaign. Discord’s ability for a user to rapidly move between audio channels, text with images, and share screens is a significant enhancement over traditional videoconferencing tools. In addition, multiple combinations of broadcast audio are made available via a web-based application to allow remote participants to simultaneously listen to general announcements/sounds while conducting their own specific conversations. Secure methodologies have been put into place to allow remote control of hardware including DIII-D’s plasma control system application. Secure methods also included the ability of the on-site team to closely coordinate their work with remote team members which has been enhanced through extensions to the wireless network and the use of tablet computers for audio/video/screen sharing. However, no amount of software can fully replace the need for ‘hands on hardware.’ This infrastructure was severely stress tested during the COVID-19 pandemic where occupancy of the DIII-D control room was restricted. Operational efficiency during the pandemic, measured in discharges per hour, remained high (3.8 ± 0.8) compared to values obtained pre-pandemic (3.7 ± 0.8).

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2022

XRF-ROI Finder: Machine Learning to Guide Region-of-Interest Scanning for X-ray Fluorescence Microscopy

The ROI-finder software is being developed for use by several Microscopy Group beamlines at Argonne National Laboratory, including 2-ID microprobes and 9-ID-B Bionanoprobe which use multi-scale scanning fluorescence microscopy to acquire elemental maps (multi-modal image data). Microscopy experiments require scan of samples at a coarse resolution followed by ROI identification using feature detection based on domain expertise. Finer resolution scans are then conducted based on identified ROI. The decision-making process based on domain expertise will be difficult to perform for faster data rates and much larger sampling volumes anticipated after APS-U necessitating the need for the ROI-finder software. The ROI- finder detects regions of interest through a continuous learning process, starting with a unsupervised representation learning and improving its recommendations through supervised learning and an interactive tool for user annotation. The scope of ongoing development efforts includes the integration of image registration module to correlate optical and X-ray images, extraction of feature morphology as well as elemental signatures in the image space and incorporation of beamtime streaming data by the scanning probe via EPICS.

CHOWDHURY, M. ARSHAD ZAHANGIR↗
DOE OSTI2023

Managing Software Provenance to Enhance Reproducibility in Computational Research

Scientific processes rely on software as an important tool for data acquisition, analysis, and discovery. Over the years, sustainable software development practices have made progress in being considered as an integral component of research. However, management of computation-based scientific studies is often left to individual researchers who design their computational experiments based on personal preferences and the nature of the study. Here, we believe that the quality, efficiency, and reproducibility of computation-based scientific research can be improved by explicitly creating an execution environment that allows researchers to provide a clear record of traceability. This is particularly relevant to complex computational studies in high-performance computing (HPC) environments. In this article, we review the documentation required to maintain a comprehensive record of HPC computational experiments for reproducibility. We also provide an overview of tools and practices that we have developed to perform such studies around Flash-X, a multiphysics scientific software.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Intelligent Experiments Through Real-time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors (Final Report)

The overall vision of this project was to integrate real-time artificial intelligence (AI) directly into the data acquisition and detector-control systems of nuclear physics experiments, including both fast online event selection and an autonomous detector-control feedback loop. The work carried out under the award focused on the fast online event-selection half of that vision: the efficient recording of low-momentum heavy-flavor (HF) hadron decays in proton-proton collisions at the sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC)—an observable that requires fast tracking and topological trigger selection not previously demonstrated at RHIC, and that is essential for QCD studies at future facilities such as the Electron-Ion Collider (EIC). The autonomous detector-control (GPU-based feedback) component named in the project title remained a design concept and was not implemented under this award. The Massachusetts Institute of Technology (MIT) group led the offline simulation and data processing needed to train the machine-learning (ML) models, the translation of trained models to Field-Programmable Gate Array (FPGA) firmware using the hls4ml framework, and the physics validation of heavy-flavor reconstruction. Over the award period, the team developed and hardware-tested the principal components of an AI-based heavy-flavor trigger on simulated and recorded sPHENIX tracker data: a software Bipartite Graph Attention Network (BiGAT) trigger model reaching > 95% signal efficiency at 99% background rejection; an FPGA-native hit clusterizer matching the offline clustering; smaller networks synthesized to FPGA within the required sub-10 µs latency; and an assembled decoder–clusterizer–inference firmware chain exercised on the FELIX readout board. A complete, fully integrated hardware demonstrator was not finished within the award period. This report documents the project goals, the MIT group’s contributions, the technical accomplishments, and the outlook toward applications at the future EIC ePIC detector.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Enabling modern data discovery for atmospheric measurements

The Atmospheric Radiation Measurement (ARM) user facility is a US Department of Energy Office of Science user facility that is managed and operated through a collaborative effort led by nine US Department of Energy national laboratories. The ARM Data Center, located at Oak Ridge National Laboratory, is responsible for the timely collection, processing, and delivery of data products to the scientific community. The ARM Data Center holds more than 11,000 data products, including metadata collected from field campaigns, instruments, value-added products, and principal investigator–contributed data. These data sets are checked for successful transfer (for most data, this transfer is carried out automatically via the network; however, some of the largest data sets and some of the most remote sites require manual shipping of hard disks) and both the data and metadata are processed to a standard format, which is an ARM-standardized structure, via the Network Common Data Form. The Network Common Data Form is a self-describing binary format with many compatible software tools. Once processed, the data are cataloged, stored in the ARM Data Archive, and made discoverable through association with an array of metadata-characterizing information, such as location and measurement classification. These metadata enable powerful search capabilities through the ARM Data Center Data Discovery interface. This paper discusses the workflow of how the new discovery system has been redesigned from user requirements and how the data are distributed to the scientific community.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Discorpy : algorithms and software for camera calibration and correction

Camera or lens-based detector calibration is essential for spatial accuracy in applications like dimensional tomography, optical metrology, and computer vision. Many methods and software exist yet there is still a lack of approaches that achieve both high accuracy and robustness while being easy to use and capable of handling a wide range of distortions. Radial lens distortion is common in high-resolution X-ray detector optics used in parallel-beam tomography at synchrotrons. Achieving sub-pixel accuracy requires calibrating with an optical target image. Although methods for characterizing radial distortion are well established, acquired images often also include perspective distortion and optical center offset. Here, we present our approaches to individually characterize and correct both types of distortion using a single calibration image, implemented in the Discorpy software.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Software-Hardware Co-design of Heterogeneous SmartNIC System for Recommendation Models Inference and Training

Deep Learning Recommendation Models (DLRMs) are critical applications in various domains and have evolved as one of the single largest machine learning applications. Trillions of DLRM parameters exceed the on-chip memory capacity of GPUs. Large-scale multi-node systems are required for distributed DLRM inference and training, which suffer from the all-to-all communication bottleneck, mainly limiting the scalability of ever-growing DLRMs. In recent years, SmartNICs have evolved with coupled computation and communication capabilities providing opportunities for a powerful heterogeneous device in the system. However, there isn't such a distributed system that fully leverages the abundant smartNIC resources that resolve the scalability issue of DLRMs. In this work, we proposed a software-hardware co-design of a heterogeneous smartNIC system that resolves the communication bottleneck of distributed DLRMs, mitigates the memory bandwidth pressure, and improves computation efficiency. We provide a set of smartNIC designs of cache systems (including local cache and remote cache) and smartNIC computation kernels which reduce data movement, relieve memory lookup intensity, and improve the GPU's computation efficiency. In addition, we propose a graph algorithm that improves the data locality of queries within batches which optimizes the overall system performance with higher data reuse. Our evaluation shows that our system achieves 2.1x latency speedup for inference and 1.6x throughput speedup for training.

Guo, Anqi↗
DOE OSTI2024

Boosting RDataFrame performance with transparent bulk event processing

RDataFrame is ROOT’s high-level interface for Python and C++ data analysis. Since it first became available, RDataFrame adoption has grown steadily and it is now poised to be a major component of analysis software pipelines for LHC Run 3 and beyond. Thanks to its design inspired by declarative programming principles, RDataFrame enables the development of highperformance, highly parallel analyses without requiring expert knowledge of multi-threading and I/O: user logic is expressed in terms of self-contained, small computation kernels tied together by a high-level API. This design completely decouples analysis logic from its actual execution, and opens several interesting avenues for workflow optimization. In particular, in this work we explore the benefits of moving internal data processing from an event-by-event to a bulkby-bulk loop. This refactoring dramatically reduces the framework’s runtime overheads; in collaboration with the I/O layer it improves data access patterns; it exposes information that optimizing compilers might use to auto-vectorize the invocation of user-defined computations; finally, while existing user-facing interfaces remain unaffected, it becomes possible to additionally offer interfaces that explicitly expose bulks of events, useful e.g. for the injection of GPU kernels into the analysis workflow. In order to inform similar future R&D, design challenges will be presented, as well as an investigation of the relevant timememory trade-off backed by novel performance benchmarks.

Guiraud, Enrico↗
DOE OSTI2025

An Open-Source Python Package for CFD Solution Verification

Informed decision-making using computational fluid dynamics (CFD) results requires quantifying the errors and uncertainties of a simulation. Verification, validation, and uncertainty quantification (VVUQ) methods were developed to address this need and have matured. However, these VVUQ analyses are often non-trivial and require CFD analysts and practitioners to have specific skill sets. This has led to the uneven adoption of VVUQ analyses, in part, based on the availability of software tools to aid CFD analysts and practitioners. Solution verification, a procedure to evaluate the accuracy of a simulation by estimating potential errors arising from the computational model and computing the uncertainties without comparing to results from a physical system, is one of the lagging VVUQ analyses as the absence of software has forced CFD analysts and practitioners to develop their own codes or piece together incomplete software from across the internet. This work presents an opensource Python package, CFDverify, to lower the barrier of entry and fill in the technological gap in solution verification. CFDverify also provides a streamlined framework to remove some potential errors in post-processing CFD results. The hope is that CFDverify can improve the quality and quantity of CFD solution verification in scientific and research studies and attract interest in developing a communal tool. This paper describes the design, features, and an example use of CFDverify.

Weinmeister, Justin [ORNL] (ORCID:0000000160090237↗
DOE OSTI2021

Laser Wavelength Shift from Splitters and Switches

The Detonation Science and Technology Group (Q-6) operates several photon Doppler velocimetry (PDV) systems as a diagnostic tool for measuring velocities in the 1–10 km/s range during high explosive (HE) detonation experiments. Q-6 uses “heterodyne” PDV systems that are produced by Traverse Solutions and Instrumentation, LLC (TS&I). These heterodyne systems combine wavelength (WL) tunable source and “reference” lasers with a WL offset of 10–100 pm from each other. Properly setting the source and reference WLs for an experiment can be critical for collecting the desired experimental data. The tunable lasers report WL readings in their control software, but these readings are based on a lookup table from a temperature reading and are not able to be calibrated. Therefore, recent test fire programs that require PDV as a diagnostic have also implemented requirements for monitoring the WL of the tunable lasers using a precision WL meter. To monitor the WL of several PDV lasers without disconnecting any patch fibers form the experimental setup, Q-6 developed a system splitters with a coupling ratio (%) of 99:1 and an optical switch with the switch output going to a single WL meter. A brief experiment was developed to determine if the splitters and/or switch introduced any WL bias into the readings. The purpose of this report is to document the system, the experiment, and its results.

42 ENGINEERING↗
DOE OSTI2021

TEMPI: An Interposed MPI Library with Canonical Representation of MPI Datatypes [Poster]

TEMPI provides a transparent non-contiguous data-handling layer compatible with various MPIs. MPI Datatypes are a powerful abstraction for allowing an MPI implementation to operate on non-contiguous data. CUDA-aware MPI implementations must also manage transfer of such data between the host system and GPU. The non-unique and recursive nature of MPI datatypes mean that providing fast GPU handling is a challenge. The same noncontiguous pattern may be described in a variety of ways, all of which should be treated equivalently by an implementation. This work introduces a novel technique to do this for strided datatypes. Methods for transferring non-contiguous data between the CPU and GPU depends on the properties of the data layout. This work shows that a simple performance model can accurately select the fastest method. Unfortunately, the combination of MPI software and system hardware available may not provide sufficient performance. The contributions of this work are deployed on OLCF Summit through an interposer library which does not require privileged access to the system to use

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

The Essence of Cryptol: A Denotational Cryptol Interpreter in Coq for Foundational Assurances for Quantum Resistant Cryptosystems

Systems of the utmost consequence need a means to establish authenticity of software and data. Cryptosystems implement authentication, but can be vulnerable to cryptographic and implementation attacks. With the threat of quantum cryptographic attacks, “post-quantum” cryptosystems (PQCs) must be henceforth used in these systems. However, the new cryptography needs new ways to, rigorously and machine-checkably, prove systems free of vulnerabilities. We propose a retargetable capability to rapidly instantiate proven correct postquantum cryptosystems through novel proof-carrying synthesis and proof-automation technique, extending those proven successful on existing systems. This capability is crucial to meeting the cryptographic requirements for future high-consequence systems. Since specifications for high consequence cryptography are presently captured in a domain specific language known as Cryptol. While this can enable convenient fully automated reasoning about Cryptol specificaitons and implementations via the Software Analysis Workbench (SAW), Cryptol has expressivity gaps, so that cryptosystems with probabilistic programming features like Falcon cannot be fully expressed in the language. Moreover, SAW’s automation fails for programs and specificaitons with inductive and recursive structure, as in the Sphincs+ PQC. Finally, Cryptol and SAW together represent some 200,000 lines of unverified Haskell, so that the any guarantees about high consequence cryptography are presently contingent on a large, unverified, yet trusted computing base. The first step of the larger project of agile, assured crpytography is therefore to provide a formal, mechanized semantics for Cryptol, so that the specifications expressed by cryptographers in Cryptol can be reasoned about and compiled into performant implementations with a foundational, machine checkable certificate of correctness. This report describes our work on this first step, culminating in the design of a certified denotational interpreter, in Coq, for core Cryptol.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Improving and Automating Building Model Data Exchange

There are many instances throughout a project’s lifecycle where there arises a need for quick and accurate risk assessment of building designs. For example, an unexpected design change during construction may necessitate structural engineers to perform a seismic risk assessment on analytical models of the updated building design using high fidelity structural analysis software, such as ANSYS or Abaqus. However, the efficiency of such workflows often depends upon the interoperability of architectural design software and structural analysis software. When the quality of this interoperability is lacking or even non-existent, the efficiency of virtual engineering workflows is hampered, which increases project costs. A McGraw Hill industry survey of professional users of Building Information Modeling (BIM) technologies found that there is high demand for BIM interoperability for structural analysis, but that the value/difficulty ratio is currently too low for practical use. There have been efforts by the academic community to facilitate model data exchange between the architectural design and structural analysis domains, but such solutions have not been widely adopted by industry, face technical challenges, and oftentimes are limited in applicability for users of various BIM software. Therefore, INL is developing capabilities to improve, automate, and generalize model data exchange between architectural BIM software (e.g., Revit) and structural analysis software (e.g., SAP2000, ANSYS). The goal is to help expedite and automate as much of the pre-processing step for creating analytical models in finite element analysis software as reasonably as possible. Such a "BIM-to-FEA" conversion tool should provide direct benefit to end-users through accuracy, automation, quick turn-around, and wide applicability. To generalize the application of this BIM-to-FEA conversion tool and increase its useability among the many different commercial BIM software currently used by industry, the program is being developed with the concept of openBIM. OpenBIM is the application of non-proprietary, open data standards that allow for BIM model data exchange in a format that is accessible, retainable, and useable for all users. The most widely used open, non-proprietary data exchange format for BIM is the Industry Foundation Classes (IFC) schema. IFC is developed by buildingSMART international and is ISO certified (ISO 16739-1:2018). The BIM-to-FEA conversion tool is being developed for compatibility with typical commercial building designs of steel framed structures. The tool is currently capable of importing architectural BIM data of framed building structures, recognizing and extracting the aspects of the model that are required for structural analysis, adjusting the connectivity of frame members, and finally exporting to an analytical model stored in the IFC format. The exported IFC analytical model can then be imported into various openBIM compliant software, such as SAP2000. Such capabilities have already been tested on commercial software, as shown above, and continue to be improved. Work is underway to test the conversion on various commercial BIM software, develop a user-friendly interface, incorporate the program into the broader DeepLynx data warehouse project being developed by INL, and to eventually open-source the tool for the benefit of the community. Future development of the tool envisions the ability for efficient iterative risk assessment of generative building designs, all within a workflow utilizing open-source tools. One such open-source tool will be MOOSE, an advanced finite element analysis tool developed at INL. The conversion tool will also branch out from typical commercial building designs and will aim to incorporate nuclear construction. The aim will be to convert both structural and non-structural components of nuclear facilities, such as curved concrete containment structures and piping systems, respectively.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

xSDK: Building an ecosystem of highly efficient math libraries for exascale

Current efforts to build increasingly powerful computer architectures are opening up new avenues for more complex and higher fidelity simulations coupled with data analytics and learning, leading to new scientific insights and deeper understanding. At one extreme, exascale computers will be much faster than previous computer generations (performing 10 18 operations per second—that is, 1,000 times faster than petascale). To achieve these performance improvements, computer architectures are becoming increasingly complex, with deep memory hierarchies, very high node and core counts, and heterogeneous features such as graphics processing units (GPUs). Such architectural changes impact the full breadth of computing scales, as heterogeneity pervades even current-generation laptops, workstations, and moderate-sized clusters. While emerging advanced architectures provide unprecedented opportunities, they also present significant challenges for developers of scientific applications, such as multiphysics and multiscale codes, who must adapt their software to handle disruptive changes in architectures and new programming models that have not yet stabilized. Developers must consider increasing concurrency while reducing communication and synchronization, and other complexities such as the potential for using mixed precision to leverage the compute power available in low-precision tensor cores. On one hand, developers must implement new scientific capabilities, which in turn increase code complexity. On the other hand, the codes must be ported to new architectures, requiring the inclusion of new programming models and the restructuring of code to achieve good performance. Addressing these issues is beyond the capability of any single person or team—leading to the need for collaboration among many teams, who encapsulate their expertise in reusable software and work together to create sustainable software ecosystems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Pynta-An Automated Workflow for Calculation of Surface and Gas-Surface Kinetics

Many important industrial processes rely on heterogeneous catalytic systems. However, given all possible catalysts and conditions of interest, it is impractical to optimize most systems experimentally. Automatically generated microkinetic models can be used to efficiently consider many catalysts and conditions. However, these microkinetic models require accurate estimation of many thermochemical and kinetic parameters. Manually calculating these parameters is tedious and error prone, involving many interconnected computations. Here, we present Pynta, a workflow software for automating the calculation of surface and gas–surface reactions. Pynta takes the reactants, products, and atom maps for the reactions of interest, generates sets of initial guesses for all species and saddle points, runs all optimizations, frequency, and IRC calculations, and computes the associated thermochemistry and rate coefficients. It is able to consider all unique adsorption configurations for both adsorbates and saddle points, allowing it to handle high index surfaces and bidentate species. Pynta implements a new saddle point guess generation method called harmonically forced saddle point searching (HFSP). HFSP defines harmonic potentials based on the optimized adsorbate geometries and which bonds are breaking and forming that allow initial placements to be optimized using the GFN1-xTB semiempirical method to create reliable saddle point guesses. This method is reaction class agnostic and fast, allowing Pynta to consider all possible adsorbate site placements efficiently. We demonstrate Pynta on 11 diverse reactions involving monodenate, bidentate, and gas-phase species, many distinct reaction classes, and both a low and a high index facet of Cu. Our results suggest that it is very important to consider reactions between adsorbates adsorbed in all unique configurations for interadsorbate group transfers and reactions on high index surfaces.

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

Accelerating Random Forest Classification on GPU and FPGA

Random Forests (RFs) are a commonly used machine learning method for classification and regression tasks spanning a variety of application domains, including bioinformatics, business analytics, and software optimization. While prior work has focused primarily on improving performance of the training of RFs, many applications, such as malware identification, cancer prediction, and banking fraud detection, require fast RF classification. In this work, we accelerate RF classification on GPU and FPGA. In order to provide efficient support for large datasets, we propose a hierarchical memory layout suitable to the GPU/FPGA memory hierarchy. We design three RF classification code variants based on that layout, and we investigate GPU- and FPGA-specific considerations for these kernels. Our experimental evaluation, performed on an Nvidia Xp GPU and on a Xilinx Alveo U250 FPGA accelerator card using publicly available datasets on the scale of millions of samples and tens of features, covers various aspects. First, we evaluate the performance benefits of our hierarchical data structure over the standard compressed sparse row (CSR) format. Second, we compare our GPU implementation with cuML, a machine learning library targeting Nvidia GPUs. Third, we explore the performance/accuracy tradeoff resulting from the use of different tree depths in the RF. Finally, we perform a comparative performance analysis of our GPU and FPGA implementations. Our evaluation shows that for high accuracy targets, our GPU implementation yields 5-9x speedup over CSR, and up to a 2x speedup over cuML.

FPGA, Xilinx FPGA, GPU, Random Forest classificati↗