Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Intel”

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 577 records · Page 32

Resilient Energy Delivery and Control Systems (REDCS) (Final Technical Report)

US critical infrastructure is increasingly the target of cyberattacks, where disturbances could cause considerable damage and disruption. To help provide a new layer of cyber-physical protection for one key energy delivery system, natural gas pipelines, GE Vernova Advanced Research along with partners Florida State University and Intel Corporation created an innovative technology called "Resilient Energy Delivery and Control Systems" (REDCS). This cybersecurity package helps detect anomalies caused by cyberattacks, isolate the subsystem being impacted by the attack, and provide functions that can allow for resiliency – giving better situational awareness to the operators and cybersecurity specialists or in the future perform closed loop control for continued operation while compromised.

03 NATURAL GAS↗

COG User's Manual: A Multiparticle Monte Carlo Transport Code (Sixth Edition)

COG is a high-resolution code for the Monte Carlo simulation of coupled particle transport in arbitrary 3-D geometry. COG will transport neutrons, protons, deuterons, alpha particles with energies up to hundreds of GeV, and photons with energy ranges limited by the available cross section sets and physics models. Electrons can be transported via the EGS5 electron transport kernel, electrons can also be transported. The COG code is a significant upgrade from earlier Monte Carlo transport codes and has been written specifically to make it more versatile, accurate, and easy to use. COG has provisions for calculating deep penetration (shielding) problems, criticality problems, and neutron activation problems while retains all of the standard capabilities found in other Monte Carlo transport codes. COG uses high-resolution pointwise cross-section databases and makes no compromises in the transport physics, so that the results of a COG run are limited only by the accuracy of the databases used. COG runs primarily on Linux Operating System workstations with MPICH software installed – currently, Red Hat 7 & 8, Windows 10 (Windows Subsystem for Linux –WSL), Ubuntu 16, 18 & 20, OpenSUSE Leap 15.2, Fedora 32, Apple Power Mac with Intel CPU (with MacPorts installed) workstations, and LLNL LC supercomputer CTS-1 cluster with TOSS 3 are supported.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Argonne Leadership Computing Facility 2024 Operational Assessment Report

This Operational Assessment Report describes how the Argonne Leadership Computing Facility (ALCF) met or exceeded every goal set by DOE for the calendar year (CY) 2024. In CY 2024, the ALCF operated Polaris, an AMD and NVIDIA-based Hewlett Packard Enterprise (HPE) Apollo 6500 Gen10+ system that provides a powerful platform for breakthrough science and prepares applications and workloads for Aurora, Argonne National Laboratory’s Intel-HPE exascale computer. Since going into production in August 2022, Polaris has supported research teams from the DOE Exascale Computing Project (ECP), which concluded in 2024, and from ALCF’s Aurora Early Science Program (ESP). Polaris delivered a total of 2555.7K node-hours to 17 Innovative and Novel Computational Impact on Theory and Experiment (INCITE) 2024 projects and 601.6K node-hours to ASCR Leadership Computing Challenge (ALCC) projects (12 of which were awarded during the 2023–2024 ALCC year and 11 of which were awarded during the 2024–2025 ALCC year), as well as providing substantial time for Director’s Discretionary (DD) projects (966.2K node-hours). As Table ES.1 shows, Polaris performed exceptionally well in terms of overall availability (97.7 percent), scheduled availability (99.6 percent), and utilization (94.7 percent; Table 2.1). In CY 2024, ALCF supported more than 2,000 users. As of March 18, 2025, ALCF’s users have published 254 papers in peer-reviewed journals and technical proceedings.

97 MATHEMATICS AND COMPUTING↗

Code Coverage Status of the ARC Code DIF3D

The Argonne Reactor Code (ARC) software system supports users in their fast reactor design goals by providing neutronic, thermal-hydraulic, and structural analysis capabilities. DIF3D plays a pivotal role in the ARC system as the primary homogenized assembly neutronic calculation methodology for fast reactor problems. Over its 40 years history, ARC software usage with DIF3D has been applied to numerous fast and thermal spectrum reactor analysis projects with good to excellent comparison against experiments. With continued improvement of computation resources, many of the geometry modeling capabilities in DIF3D that were primarily used in low order schemes are not really needed anymore. Today, the diffusion and transport capabilities of DIF3D-VARIANT are primarily used in the reactor design process with some scattered usage of DIF3D-FD and DIF3D-Nodal. In recent work, the DIF3D software verification was completed for DIF3D-FD and DIF3D-VARIANT on the geometry options used in the Versatile Test Reactor project. While we can be confident that these capabilities of DIF3D are well used and thus trusted, it does not demonstrate that all possible input options of DIF3D are actually working, but just those that were tested as part of VTR are and that they are correct. Thus, the purpose of the present work is to identify a set of test problems for DIF3D and assess the code coverage of DIF3D for those test problems. The goal is to document what parts of the existing DIF3D code are touched by the set of test problems and which are not. Because the verification work done on DIF3D-VARIANT and DIF3D-FD was focused on the most common uses of DIF3D for fast reactor analysis, the code coverage assessment of those capabilities is the highest priority. This will ensure that nothing is being missed by the existing verification test problems that DIF3D relies upon. The DIF3D-Nodal capability will also be inspected for code coverage as part of this work to further ensure that regular regression testing of DIF3D will trap any likely errors the end user might experience with the DIF3D software. The code coverage analysis of DIF3D was performed with the Code Coverage Tool of the Intel Fortran compiler which requires modifications to the compilation of DIF3D. The detailed coverage tables are given for each submodule of DIF3D separately, and for the submodules which are primarily developed for DIF3D, most of the source files could be at least partially touched. Most of the uncovered parts/files could be easily ignored, because they are either for error message and debugging output or obviously not needed by DIF3D. Out of the entire source codes of DIF3D, only a few uncovered modules deserve further investigation.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Integrating Machine-learning-assisted Computer Vision with RICH System

Developments in artificial intelligence have vastly expanded the capabilities of robots. Currently, the Spallation Neutron Source (SNS) beamlines at Oak Ridge National Lab (ORNL) have robotic sample loaders to increase the efficiency of running experiments. However, they require retraining if anything about the situation changes, e.g., where the samples are, and cannot notice if errors occur. So, the viability of using computer vision and machine learning to enhance these sample loaders’ functionality was investigated. In this project, the RICH system with a Dobot CR3 6-axis robot present at the VULCAN beamline assisted by an Intel Realsense D435i camera, a unique camera that enables convenient translation of 2D pixel coordinates to 3D world points, was programmed to load ceramic crucibles into a thermogravimetric analyzer (TGA) furnace. An algorithm was constructed in Python with three major phases planned: (1) obtaining a sample, (2) moving it to the target location, and then (3) bringing the sample back to its original location once the experiment finished. In the first phase, the algorithm would dynamically detect sample locations using ArUco markers to recognize the samples’ general location and a custom-trained yolov5 object detection model to locate the crucibles’ centers. Afterward, the robot would be directed to pick up samples based on the crucibles’ calculated positions. In the second phase, the robot would move the sample to a secondary point, reorient its grip, and place the sample at the target location. In the final phase, the robot would determine whether the sample was intact and would bring it back to its original place if it was or raise an alarm. Using this algorithm, the robot was able to pick up different types of crucibles at varying positions. These results indicate that integrating machine-learning-assisted computer vision with robotic sample loaders can result in effective autonomous detection of samples.

97 MATHEMATICS AND COMPUTING↗

Code Coverage Status of the ARC Code RCT

The Argonne Reactor Code (ARC) software system supports users in their fast reactor design goals by providing neutronic, thermal-hydraulic, and structural analysis capabilities. REBUS plays a pivotal role in the ARC system as the primary fuel cycle analysis capability for fast reactor problems. Over its 60 year history, ARC software usage with REBUS has been applied to numerous fast and thermal spectrum reactor analysis projects with good to excellent comparison against experiments. The RCT code is a later addition and uses the REBUS restart files to define its input. The RCT code was built to provide pin depletion details on EBR-II models and thus many features of RCT were specifically tailored to the needs of EBR-II models. Additional approximations were invoked which are likely only valid for the EBR-II reactor and the particular fuel management that was done for it. The purpose of the present work is to identify a set of test problems for RCT and assess the code coverage for those test problems. The goal is to document what parts of the existing RCT code are touched by the set of test problems and which are not. Because no detailed verification work has been done on RCT, the existing regression testing suite was chosen for the code coverage assessment. The code coverage analysis of RCT was performed with the Code Coverage Tool of the Intel Fortran compiler which requires modifications to the compilation of RCT. The detailed coverage tables are given for each part of RCT. As will be discussed and shown, some parts of the RCT capability that are known to be used by the EBR-II analysis work are not tested by the regression testing suite. These aspects should be resolved before major source code changes are taken for the RCT software. Because REBUS and DIF3D are not subroutines of RCT, the coverage changes in both of those codes is not altered by RCT. The same is true for all of the modules of DIF3D that are used by RCT such as SYSLIB and SEGLIB.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Code Coverage Status of the ARC Code PERSENT

The Argonne Reactor Code (ARC) software system supports users in their fast reactor design goals by providing neutronic, thermal-hydraulic, and structural analysis capabilities. PERSENT fulfills the role of generating reactivity coefficients for a given time point of a REBUS calculation usable in a point kinetics based safety analysis capability. PERSENT also provides a sensitivity coefficient capability on eigenvalue, reactivity worth, and several other key coefficients that are used in the follow-on safety analysis. Given a co-variance matrix, PERSENT can carry out the uncertainty quantification to indicate the amount of error in the reactivity coefficients derived from the errors in the cross section measurements. With continued improvement of computational resources, many of the geometry modeling capabilities in DIF3D that were primarily used in low order schemes are not really needed anymore. Today, the diffusion and transport capabilities of DIF3D-VARIANT are primarily used in the reactor design process with some scattered usage of DIF3D-FD and DIF3D-Nodal. PERSENT is part of the ARC code system and is built around DIF3D-VARIANT and the flux solution it provides. The purpose of the present work is to identify a set of test problems for PERSENT and assess the code coverage of PERSENT for those test problems. PERSENT treats the DIF3D executable as an external executable and thus the code coverage considerations only need to focus on the PERSENT source code and only a fraction of the connected modules in the existing ARC software library. The goal is to document what parts of the existing PERSENT code are touched by the set of test problems and which are not. Because the verification work done on PERSENT was focused on the most common uses of PERSENT for fast reactor analysis, the code coverage assessment of those capabilities is the highest priority. This will ensure that nothing is being missed by the existing verification test problems that users of PERSENT rely upon. The code coverage analysis of PERSENT was performed with the Code Coverage Tool of the Intel Fortran compiler which requires modifications to the compilation of PERSENT. The detailed coverage tables are given for each submodule of PERSENT. Most of the uncovered parts/files could be easily ignored because they are either for error message and debugging output or not needed by PERSENT today. Only a few uncovered parts of PERSENT deserve extending the verification test suite.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Code Coverage Status of the ARC code GAMSRC

The Argonne Reactor Code (ARC) software system supports users in their fast reactor design goals by providing neutronic, thermal-hydraulic, and structural analysis capabilities. GAMSOR serves as procedure to obtain the neutron and gamma power distribution information in the ARC code system. GAMSOR is a specially modified version of DIF3D (dif3d.x becomes dif3d_gamsor.x). After some work, it was determined that carrying out software on GAMSOR was impractical and would not fit well with commercial grade dedication. GAMSRC was written in the last decade to replace GAMSOR such that the user can rely soley upon the verified DIF3D (dif3d.x) code and GAMSRC to obtain the neutron and gamma power distribution information. GAMSRC is also fully verified and ready for commercial grade dedication. In the coming years, GAMSOR will be deprecated and GAMSRC will fully take over in the ARC code system. This document identifies the set of test problems used to assess the code coverage for GAMSRC. The goal is to document what parts of the existing GAMSRC code are touched by the set of test problems and which are not. The code coverage analysis of GAMSRC was performed with the Code Coverage Tool of the Intel Fortran compiler which requires modifications to the compilation of GAMSRC. The code coverage tables are given for each submodule of GAMSRC. Because GAMSRC links to modules in DIF3D, some details on coverage changes to the DIF3D lined files is provided. As will be seen, most of the uncovered parts/files can be ignored because they are either for error message and debugging output or obviously not needed by GAMSRC today.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Code Coverage Status of the ARC Code DASSH-F

The Argonne Reactor Code (ARC) software system supports users in their fast reactor design goals by providing neutronic, thermal-hydraulic, and structural analysis capabilities. DASSH-F serves as a steady state thermal hydraulic capability within the ARC system and replaces the SE2-ANL software that preceded it. This document identifies the set of test problems used to assess the code coverage for DASSH-F. The goal is to document what parts of the existing DASSH-F code are touched by the set of test problems and which are not. Because the verification work remains to be done on DASSH-F, one can assume that most of these issues will be resolved as part of that work. The code coverage analysis of DASSH-F was performed with the Code Coverage Tool of the Intel Fortran compiler which requires modifications to the compilation of DASSH-F. The code coverage tables are given for each submodule of DASSH-F. Because DASSH-F links to modules in DIF3D, some details on coverage changes to the DIF3D linked files is provided. As will be seen, most of the uncovered parts/files can be ignored because they are either for error message and debugging output or obviously not needed by DASSH-F today. Seven features of the DASSH-F code were identified to not be covered by the existing testing suite and thus additional verification test problems are suggested to fully cover these sections.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

MPACT Development Activities at University of Michigan

This report documents the efforts of general development tasks in MPACT performed primarily at the University of Michigan during FY17. These tasks included: 1. Unstructured CMFD - The unstructured CMFD supports the RIA and BWR challenge problems. It allows MPACT to simulate mixed lattice cores (e.g. 8x8 next to 9x9). This design feature is common to operating BWRs and also the SPERT experiments being used to validate VERA-CS for RIA. 2. Higher order spatial sources for MOC - The higher order sources are for improving the overall performance of MPACT. Higher order spatial sources in MOC provide accurate solutions with potentially significantly fewer mesh regions, thus speeding up every calculation. This method is also particularly important for BWRs. 3. Improving fine source region visualization - The improvement of the visualization of the flat source region mesh in MPACT is to support users and communication of models and results from MPACT. VERAView provides visualization of pin-wise information, however MPACT resolves the spatial solution to intra pin level. The capability provided here will let users visualize the full model geometry and components of the solution at the highest spatial resolution available from MPACT. 4. Automated spatial decomposition - The automated spatial decomposition removes the burden from the user to know how to spatially decompose the model for MPACT; a task that has been rather tricky in the past. Instead the user may give some number of processors that is a multiple of the axial mesh and MPACT will determine a nearly optimal decomposition that gives better load balance than most decompositions used currently. 5. Optimal relaxation factors for feedback - The development of optimal relaxation factor is to improve the stability and robustness of problems with feedback. Presently the default feedback values are 0.5, which may or may not be optimal. Without relaxation the fixed point iteration that is currently used is known to be unstable for large problems. The optimal relaxation factors are derived by utilizing a previous Fourier analysis. From the original milestone there was a scope change where the work on the Intel MIC was deferred for work on the automated spatial decomposition.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Developments in Performance and Portability for MadGraph5_aMC@NLO

Event generators simulate particle interactions using Monte Carlo techniques, providing the primary connection between experiment and theory in experimental high energy physics. These software packages, which are the first step in the simulation worflow of collider experiments, represent approximately 5 to 20% of the annual WLCG usage for the ATLAS and CMS experiments. With computing architectures becoming more heterogeneous, it is important to ensure that these key software frameworks can be run on future systems, large and small. In this contribution, recent progress on porting and speeding up the Madgraph5_aMC@NLO event generator on hybrid architectures, i.e. CPU with GPU accelerators, is discussed. The main focus of this work has been in the calculation of scattering amplitudes and "matrix elements", which is the computational bottleneck of an event generation application. For physics processes limited to QCD leading order, the code generation toolkit has been expanded to produce matrix element calculations using C++ vector instructions on CPUs and using CUDA for NVidia GPUs, as well as using Alpaka, Kokkos and SYCL for multiple CPU and GPU architectures. Performance is reported in terms of matrix element calculations per time on NVidia, Intel, and AMD devices. The status and outlook for the integration of this work into a production release usable by the LHC experiments, with the same functionalities and very similar user interfaces as the current Fortran version, is also described.

Valassi, Andrea↗

Progressive CPU Volume Rendering with Sample Accumulation

Here, we present a new method for progressive volume rendering by accumulating object-space samples over successively rendered frames. Existing methods for progressive refinement either use image space methods or average pixels over frames, which can blur features or integrate incorrectly with respect to depth. Our approach stores samples along each ray, accumulates new samples each frame into a buffer, and progressively interleaves and integrates these samples. Though this process requires additional memory, it ensures interactivity and is well suited for CPU architectures with large memory and cache. This approach also extends well to distributed rendering in cluster environments. We implement this technique in Intel’s open source OSPRay CPU ray tracing framework and demonstrate that it is particularly useful for rendering volumetric data with costly sampling functions.

97 MATHEMATICS AND COMPUTING↗

FPGA Acceleration of GCN in Light of the Symmetry of Graph Adjacency Matrix

Graph Convolutional Neural Networks (GCNs) are widely used to process large-scale graph data. Different from deep neural networks (DNNs), GCNs are sparse, irregular, and unstructured, posing unique challenges to hardware acceleration with regular processing elements (PEs). In particular, the adjacency matrix of a GCN is extremely sparse, leading to frequent but irregular memory access, low spatial/temporal data locality and poor data reuse. Furthermore, a realistic graph usually consists of unstructured data (e.g., unbalanced distributions), creating significantly different processing times and imbalanced workload for each node in GCN acceleration. To overcome these challenges, we propose an end-to-end hardware-software co-design to accelerate GCNs on resource-constrained FPGAs with the features including: (1) A custom dataflow that leverages symmetry along the diagonal of the adjacency matrix to accelerate feature aggregation for undirected graphs. We utilize either the upper or the lower triangular matrix of the adjacency matrix to perform aggregation in GCN to improve data reuse. (2) Unified compute cores for both aggregation and transform phases, with full support to the symmetry-based dataflow. These cores can be dynamically reconfigured to the systolic mode for transformation or as individual accumulators for aggregation in GCN processing. (3) Preprocessing of the graph in software to rearrange the edges and features to match the custom dataflow. This step improves the regularity in memory access and data reuse in the aggregation phase. Moreover, we quantize the GCN precision from FP32 to INT8 to reduce the memory footprint without losing the inference accuracy. We implement our accelerator design in Intel Stratix10 MX FPGA board with HBM2, and demonstrate 1.3x-110.5x improvement in end-to-end GCN latency as compared to the state-of the-art FPGA implementations, on the graph datasets of Cora, Pubmed, Citeseer and Reddit.

Nair, Gopikrishnan R.↗

OctoFAS: A Two-Level Fair Scheduler That Increases Fairness in Network-Based Key-Value Storage

We identified a fairness problem in a network-based key-value storage system using Intel Storage Performance Development Kit (SPDK) in a multitenant environment. In such an environment, each tenant’s I/O service rate is not fairly guaranteed compared to that of other tenants. To address the fairness problem, we propose OctoFAS, a two-level fair scheduler designed to improve overall throughput and fairness among tenants. The two-level scheduler of OctoFAS consists of (i) inter-core scheduling and (ii) intra-core scheduling. Through inter-core scheduling, OctoFAS addresses the load imbalance problem that is inherent in SPDK on the storage server by dynamically migrating I/O requests from overloaded cores to underloaded cores, thereby increasing overall throughput. Intra-core scheduling prioritizes handling requests from starving tenants over well-fed tenants within core-specific event queues to ensure fair I/O services among multiple tenants. OctoFAS is deployed on a Linux cluster with SPDK. Through extensive evaluations, we found that OctoFAS ensures that the total system throughput remains high and balanced, while enhancing fairness by approximately 10% compared to the baseline, when both scheduling levels operate in a hybrid fashion.

97 MATHEMATICS AND COMPUTING↗

Cleaner Chips: Decarbonization in Semiconductor Manufacturing

The growth of the information and communication technology sector has vastly accelerated in recent decades because of advancements in digitalization and Artificial Intelligence (AI). Scope 1, 2, and 3 greenhouse gas emissions data of the top six semiconductor manufacturing companies (Samsung Electronics, Taiwan Semiconductor Manufacturing Corporation, Micron, SK Hynix, Kioxia, and Intel) were gathered from the publicly accessible Carbon Disclosure Project’s (CDP) website for 2020. Scope 3 emissions had the largest share in total annual emissions with an average share of 52%, followed by Scope 2 (32%) and Scope 1 (16%). Because of the absence of a standardized methodology for Scope 3 emissions estimation, each company used different methodologies that resulted in differences in emissions values. An analysis of the CDP reporting data did not reveal information on strategies implemented by companies to reduce Scope 3 emissions. The use of renewable energy certificates had the largest effect on decarbonization centered on reducing Scope 2 emissions, followed by the deployment of perfluorocarbon reduction technologies to help reduce Scope 1 fugitive emissions. Technology-specific marginal abatement costs of CO2 were also estimated and varied between −416 and 12,215 USD/t CO2 eq., which primarily varied depending on the technology deployed.

42 ENGINEERING↗

Radio-flaring Ultracool Dwarf Population Synthesis

Over a dozen ultracool dwarfs (UCDs), low-mass objects of spectral types ≥M7, are known to be sources of radio flares. These typically several-minutes-long radio bursts can be up to 100% circularly polarized and have high brightness temperatures, consistent with coherent emission via the electron cyclotron maser operating in approximately kilogauss magnetic fields. Recently, the statistical properties of the bulk physical parameters that describe these UCDs have become described adequately enough to permit synthesis of the population of radio-flaring objects. For the first time, I construct a Monte Carlo simulator to model the population of these radio-flaring UCDs. This simulator is powered by Intel Secure Key (ISK), a new processor technology that uses a local entropy source to improve random number generation that has heretofore been used to improve cryptography. The results from this simulator indicate that only ∼5% of radio-flaring UCDs within the local interstellar neighborhood (<25 pc away) have been discovered. I discuss a number of scenarios that may explain this radio-flaring fraction and suggest that the observed behavior is likely a result of several factors. The performance of ISK as compared to other pseudorandom number generators is also evaluated, and its potential utility for other astrophysical codes is briefly described.

79 ASTRONOMY AND ASTROPHYSICS↗

Entity—Hardware-agnostic Particle-in-cell Code for Plasma Astrophysics. I. Curvilinear Special Relativistic Module

Entity is a new-generation, fully open-source particle-in-cell (PIC) code developed to overcome key limitations in astrophysical plasma modeling, particularly the extreme separation of scales and the performance challenges associated with evolving, GPU-centric computing infrastructures. It achieves hardware-agnostic performance portability across various GPU and CPU architectures using the Kokkos library. Crucially, Entity maintains a high standard for usability, clarity, and customizability, offering a robust and easy-to-use framework for developing new algorithms and grid geometries, which allows extensive control without requiring edits to the core source code. This paper details the core general-coordinate special relativistic module. Entity is the first PIC code designed to solve the Vlasov–Maxwell system in general coordinates, enabling a coordinate-agnostic framework that provides the foundational structure for straightforward extension to arbitrary coordinate geometries. The core methodology achieves numerical stability by solving particle equations of motion in the global orthonormal Cartesian basis, despite using generalized coordinates like Cartesian, axisymmetric spherical, and quasi-spherical grids. Charge conservation is ensured via a specialized current deposition technique using conformal currents. The code exhibits robust scalability and performance portability on major GPU platforms (AMD MI250X, NVIDIA A100, and Intel Max Series), with the 3D particle pusher and the current deposition operating efficiently at about 2 ns per particle per time step. Functionality is validated through a comprehensive suite of standard Cartesian plasma tests and the accurate modeling of relativistic magnetospheres in curvilinear axisymmetric geometries.

Hakobyan, Hayk [Flatiron Institute, New York, NY (↗

ML-based Real-Time Control at the Edge: An Approach Using hls4ml

This study focuses on implementing a real-time control system for a particle accelerator facility that performs high energy physics experiments. A critical operating parameter in this facility is beam loss, which is the fraction of particles deviating from the accelerated proton beam into a cascade of secondary particles. Accelerators employ a large number of sensors to monitor beam loss. The data from these sensors is monitored by human operators who predict the relative contribution of different sub-systems to the beam loss. Using this information, they engage control interventions. In this paper, we present a controller to track this phenomenon in real-time using edge-Machine Learning (ML) and support control with low latency and high accuracy. We implemented this system on an Intel Arria 10 SoC. Optimizations at the algorithm, high-level synthesis, and interface levels to improve latency and resource usage are presented. Our design implements a neural network, which can predict the main source of beam loss (between two possible causes) at speeds up to 575 frames per second (fps) (average latency of 1.74 ms). The practical deployed system is required to operate at 320 fps, with a 3ms latency requirement, which has been met by our design successfully.

Shi, R.↗