Processing Unit Research Articles

Novel GPU-Based Method for the Generalized Maximum Flow Problem

This paper investigates the application of a minimum loss path finding algorithm to determine the maximum flow in generalized networks that are characterized by arc losses or gains. In these generalized network flow problems, each arc has not only a defined capacity but also a loss or gain factor, which must be taken into consideration when calculating the maximum achievable flow. This extension of the traditional maximum flow problem requires a more comprehensive approach, where the maximum amount of flow is determined by accounting for additional factors such as costs, varying arc capacities, and the specific loss or gain associated with each arc. This paper extends the classic Ford–Fulkerson algorithm, adapting it to iteratively identify source-to-sink (s − t) residual directed paths with minimum cumulative loss and generalized augmenting paths (GAPs), thus enabling the efficient computation of maximum flow in such complex networks. Moreover, to enhance the computational performance of the proposed algorithm, we conducted extensive studies on parallelization techniques using graphics processing units (GPUs). Significant improvements in the algorithm’s efficiency and scalability were achieved. The results demonstrate the potential of GPU-accelerated computations in handling real-world applications where generalized network flows with arc losses and gains are prevalent, such as in telecommunications, transportation, or logistics networks.

Design, Development and Optimization of a Robotic Weed Control System for Greenhouses

This study presents the design and development of a robotic prototype specifically engineered for weed elimination within cucumber plants growing in greenhouse conditions. Given the limited research in automated weed control tailored to cucumber cultivation, this project fills a crucial gap in agricultural robotics. The cucumber plants are arranged in a single row, maintaining a spacing of approximately 40 cm between each plant, while the rows themselves are positioned about 1 meter apart. The greenhouse environment features sandy loam soil, which informs the robot's design and operational parameters. The robot employs three arrays of ultrasonic sensors strategically positioned to enhance weed detection capabilities. A PIC18 F4550-E/P microcontroller serves as the processing unit, enabling real-time feedback from the sensors that triggers the actuation of a robotic arm. This arm is engineered to maneuver between plant rows, utilizing rotating blades for efficient weed cutting. A comprehensive experimental procedure was conducted, comprising 54 trials within the greenhouse setting, to optimize the performance of the robotic arm. Key variables investigated included arm motor (AM) speed, blade rotation (BR) speed, and blade design. The research tested three BR speeds—3500, 2500, and 1500 RPM—determined based on prior literature concerning rotary movers. Furthermore, two AM speeds (10 and 30 RPM) paired with three distinct blade designs—S-shaped, triangular, and circular—were evaluated primarily for their cost-efficiency and performance based on initial testing outcomes. Results revealed a significant correlation between motor speed and weed cutting efficiency, showing that higher motor speeds correlated with a decreased percentage of cut weeds. Consequently, the 10 RPM motor speed was identified as the optimal setting for arm movement, balancing effective weed removal with operational efficiency.

Laser-Micro-Annealing of Microcrystalline Ni-Rich NCM Oxide: Towards Micro-Cathodes Integrated on Polyethylene Terephthalate Flexible Substrates

Here in this work, we report on micro-Raman spectroscopy investigations performed on freestanding Ni-rich NCM (LixNi0.83Co0.11Mn0.06O2) microcrystals transferred to flexible polyethylene terephthalate (PET) host substrates. This technological procedure introduces a first building block for future on-chip-integrated micro-accumulators for applications in flexible optoelectronics, sensors, microbiology, and human medicine. An after-synthesis thermal treatment was used to help improve the material homogeneity and perfection of the cathode material. To this end, a local laser micro-annealing process was applied to the freestanding Ni-rich NCM microcrystals. The thermally initialized structural processes in the singular micro-cathode units were characterized and determined by micro-Raman spectroscopy. Micro-Raman mapping images revealed the evolution of a recrystallization process after the local annealing procedure. Furthermore, laser micro-annealing led to the suppression of the pristine “polycrystalline morphology” of the investigated micro-cathode regions. Besides the dominant characteristic Raman mode at ~1085 cm−1, most likely ascribed to lithium carbonate, metal oxides with Raman modes around ~550 cm−1 were identified. This highly efficient transfer and integration technology represents a basic building block towards micrometer-sized accumulators for a large range of emerging applications.

Seismic full-waveform inversion regularized with a migration image

Seismic full-waveform inversion (FWI) aims to reconstruct high-resolution velocity models of the subsurface from seismic data. However, it is known to face the problem of ill-posedness and has difficulties in reconstructing deep, low-illumination regions. To overcome these challenges, we use a migration image to regularize the FWI problem. In contrast to conventional approaches where the model regularization is added to the data misfit term, our method uses a U-net to introduce this prior information. We show that the prior information about the structure contained in a migrated image can be naturally and effectively incorporated into the generated output, and therefore use the migration image as input to the U-net. We reparametrize the velocity model as a combination of the U-net output and a background velocity model. The learnable parameters of the U-net are optimized against the FWI data misfit by using the gradient calculated from automatic differentiation. Our method requires no network pre-training and is time efficient due to the use of automatic differentiation accelerated by Graphic Processing Units (GPUs). We implement this method on the synthetic Marmousi and Sigsbee2a models, as well as on field data from the Gulf of Mexico. The results show a regularization effect for mitigating local minima issues, improving convergence speed, and preserving geological structures. Such a regularization effect is well suited to improve velocity reconstruction under poor illumination conditions, such as subsalt structures.

ABFML: A problem-oriented package for rapidly creating, screening, and optimizing new machine learning force fields.

Machine Learning Force Fields (MLFFs) require ongoing improvement and innovation to effectively address challenges across various domains. Developing MLFF models typically involves extensive screening, tuning, and iterative testing. However, existing packages based on a single mature descriptor or model are unsuitable for this process. Therefore, we developed a package named ABFML, based on PyTorch, which aims to promote MLFF innovation by providing developers with a rapid, efficient, and user-friendly tool for constructing, screening, and validating new force field models. Moreover, by leveraging standardized module operations and cutting-edge machine learning frameworks, developers can swiftly establish models. In addition, the platform can seamlessly transition to the graphics processing unit environments, enabling accelerated calculations and large-scale parallel simulations of molecular dynamics. In contrast to traditional from-scratch approaches for MLFF development, ABFML significantly lowers the barriers to developing force field models, thereby expediting innovation and application within the MLFF development domains.

Scalable Networking of Neutral-Atom Qubits: Nanofiber-Based Approach for Multiprocessor Fault-Tolerant Quantum Computers

Neutral atoms are among the leading platforms toward realizing fault-tolerant quantum computation (FTQC). However, scaling up a single neutral-atom device beyond 104 atoms to meet the demands of FTQC for practical applications remains a challenge. To overcome this challenge, we clarify the criteria and technological requirements for further scaling based on multiple neutral atom quantum processing units (QPUs) connected through photonic networking links. Our quantitative analysis shows that nanofiber optical cavities have the potential as an efficient atom-photon interface to enable fast entanglement generation between atoms in distinct neutral-atom modules, allowing multiple neutral-atom QPUs to operate cooperatively without sacrificing computational speed. Using state-of-the-art millimeter-scale nanofiber cavities with the finesse of thousands, over a hundred atoms can be coupled to the cavity mode with an optical tweezer array, with expected single-atom cooperativity exceeding 100 for telecom-band transition of ytterbium atoms. This enables efficient time-multiplexed entanglement generation with a predicted Bell-pair generation rate of 105s−1 while maintaining a small footprint for channel multiplexing. These proposals and results indicate a promising pathway for building large-scale multiprocessor fault-tolerant quantum computers using neutral atoms, nanofiber optical cavities, and fiber-optic networks. Published by the American Physical Society 2025

Transition to GPU-based reconstruction for clinical organ-targeted PET scanner.

This article explores a new Graphics Processing Unit (GPU)-based techniques for efficient image reconstruction in organ-targeted Positron Emission Tomography (PET) scanners with planar detectors .
Approach: GPU-based reconstruction is applied to the Radialis low-dose organ-targeted PET technology, developed to overcome the issues of high exposure and limited spatial resolution inherent in traditional whole-body PET/CT (Computed Tomography) scans. The Radialis planar detector technology is based on four-side tileable sensor modules that can be seamlessly combined into a sensing area of the needed size, optimizing the axial field-of-view (AFOV) for specific organs, and maximizing geometric sensitivity. The article explores the transition from Central Processing Unit (CPU)-based Maximum Likelihood Expectation Maximization (MLEM) algorithms to a GPU-based counterpart, demonstrating a tenfold overall speedup in image reconstruction with a hundredfold improvement in iteration speed.
Main Results: Through standardized PET performance tests and clinical image analysis, this work demonstrates that GPU-based image reconstruction maintains diagnostic image quality while significantly reducing reconstruction times. The application of this technology, particularly in breast imaging using the Radialis Low-Dose Positron Emission Mammography (LD-PEM), significantly reduces exam times thus improving patient comfort and throughput in clinical settings.
Significance: This study represents an important advancement in the clinical workflow of PET imaging, providing insights into optimizing reconstruction algorithms to effectively leverage the parallel processing capabilities of GPUs.&#xD.

Hardware-assisted virtualization extensions for LEON processors in mixed-criticality systems

The increasing complexity of real-time embedded critical systems has driven the adoption of new methodologies to mitigate high development costs. One of the most common approaches is the implementation of mixed-criticality systems, characterized by integrating applications with different levels of criticality on the same processing unit. In these systems, applications run on a separation kernel hypervisor, a software element that controls the execution of the different operating systems, providing a virtualized environment and ensuring the necessary spatial and temporal isolation. This paper presents the design and implementation of hardware virtualization extensions for LEON processors, whose use is widespread in the field of space systems. These extensions enable the execution of virtualized applications with minimal transitions to the hypervisor, enhancing system performance. Our proposed solution defines a specific execution mode and duplicates control and status registers for the exclusive use of virtualized applications. In addition, the functionality of the hardware and software interrupt signals has been extended, allowing developers to select which ones are handled by the hypervisor and which ones by the guest operating systems directly. We have implemented the proposed extension using the LEON version 3 processor’s original VHDL code, and validated it using exhaustive tests to evaluate performance and resource consumption. The results show that the proposed modifications allow virtualized applications to execute without hypervisor intervention, matching the performance when non-virtualized while significantly outperforming existing para-virtualization solutions. Resource consumption increases by 6% to 14%, depending on the FPGA, which is low when compared to available resources. Power consumption increases by only a few milliwatts, which can be considered negligible.

Multithreaded and GPU-Based Implementations of a Modified Particle Swarm Optimization Algorithm with Application to Solving Large-Scale Systems of Nonlinear Equations

This paper presents a novel Graphics Processing Unit (GPU) accelerated implementation of a modified Particle Swarm Optimization (PSO) algorithm specifically designed to solve large-scale Systems of Nonlinear Equations (SNEs). The proposed GPU-based parallel version of the PSO algorithm uses the inherent parallelism of modern hardware architectures. Its performance is compared against both sequential and multithreaded Central Processing Unit (CPU) implementations. The primary objective is to evaluate the efficiency and scalability of PSO across different hardware platforms with a focus on solving large-scale SNEs involving thousands of equations and variables. The GPU-parallelized and multithreaded versions of the algorithm were implemented in the Julia programming language. Performance analyses were conducted on an NVIDIA A100 GPU and an AMD EPYC 7643 CPU. The tests utilized a set of challenging, scalable SNEs with dimensions ranging from 1000 to 5000. Results demonstrate that the GPU accelerated modified PSO substantially outperforms its CPU counterparts, achieving substantial speedups and consistently surpassing the highly optimized multithreaded CPU implementation in terms of computation time and scalability as the problem size increases. Therefore, this work evaluates the trade-offs between different hardware platforms and underscores the potential of GPU-based parallelism for accelerating SNE solvers.

Optimal task partitioning to minimize failure in heterogeneous computational platform

The increased energy consumption by heterogeneous cloud platforms surges the carbon emissions and reduces system reliability, thus, making workload scheduling an extremely challenging process. The dynamic voltage- frequency scaling (DVFS) technique provides an efficient mechanism in improving the energy efficiency of cloud platform; however, employing DVFS reduces reliability and increases the failure rate of resource scheduling. Most of the current workload scheduling methods have failed to optimize the energy and reliability together under a central processing unit - graphical processing unit (CPU-GPU) heterogeneous computing platform; As a result, reducing energy consumption and task failure are prime issues this work aims to address. This work introduces task failure minimization (TFM) through optimal task partitioning (OTP) for workload scheduling in the CPU-GPU cloud computational platform. The TFM-OTP introduces a task partitioning model for the CPU-GPU pair; then, it provides a DVFS- based energy consumption model. Finally, the energy-load optimization problem is defined, and the optimal resource allocation design is presented. The experiment is conducted on two standard workloads namely SIPHT and CyberShake workload. The result shows that the proposed TFA-OTP model reduces energy consumption by 30.35%, reduces makespan by 70.78% and reduces task failure energy overhead by 83.7% in comparison with energy minimized scheduling (EMS) approach.

Enhancing automatic license plate recognition in Indian scenarios

Automatic license plate recognition (ALPR) technology has gained widespread use in many countries, including India. With the explosion in the number of vehicles plying over the roads in the past few years, automating the process of documenting vehicle license plates for use by law enforcement agencies and traffic management authorities has great significance. There have been various advancements in the object detection, object tracking, and optical character recognition domain but integrated pipelines for ALPR in Indian scenarios are a rare occurrence. This paper proposes an architecture that can track vehicles across multiple frames, detect number plates and perform optical character recognition (OCR) on them. A dataset consisting of Indian vehicles for the detection of oblique license plates is collected and a framework to increase the accuracy of OCR using the data across multiple frames is proposed. The proposed system can record license plate readings of vehicles averaging 527.99 and 2157.09 ms per frame using graphics processing unit (GPU) and central processing unit (CPU) respectively.

Fast-Track Extubation Protocol for Adult Cardiac Surgery Patients to Reduce Intubation Times and Length of Stay in the Intensive Care Unit.

Prolonged intubation has been associated with unfavorable outcomes after cardiac surgery. A standardized approach is needed to ensure prompt extubation and shorten intensive care unit stays. This quality improvement project was designed to evaluate the impact of a fast-track extubation protocol on time to extubation and intensive care unit length of stay. The intervention group consisted of 26 adult cardiac surgery patients who underwent the fast-track extubation protocol. A Mann-Whitney test was used to compare time to extubation and intensive care unit length of stay in this group with those of a pair-matched control group of patients from the previous year who did not undergo the fast-track extubation protocol. An evidence-based literature review was used to develop a fast-track extubation protocol involving extubation in less than 6 hours. An educational activity was created to improve intensive care unit staff members' knowledge of the fast-track extubation protocol, and its effectiveness was measured by a posttest score of 80%. The percentage of patients with extubation times of less than 6 hours was significantly higher in the fast-track extubation protocol group than in the pair-matched control group (U = 179, P = .003). The mean intensive care unit stay decreased from 2.92 days in the control group to 1.85 days in the fast-track extubation protocol group. Implementing a fast-track extubation protocol for adult cardiac surgery patients shortened time to extubation and intensive care unit stay, expediting and improving recovery processes in the intensive care unit.

Graphics Processing Unit Simulation of Magnetic Domains Under Alternating Field Excitation Including Iron Enhancement

Graphics Processing Unit Simulation of Magnetic Domains Under Alternating Field Excitation Including Iron Enhancement

Graphics processing unit accelerated modeling of thrombus formation in cardiovascular systems using smoothed particle hydrodynamics

Cardiovascular diseases remain a global health threat, often due to uncontrolled thrombus formation. Understanding its biochemical, biological, and mechanical aspects is essential. Given the challenges of in-vivo studies, computational fluid dynamics has emerged as a cost-effective alternative. This research introduces a novel methodology for modeling thrombus formation and its growth, utilizing smoothed particle hydrodynamics (SPH). The approach is optimized for execution on graphics processing unit, significantly reducing the runtime of time-intensive thrombus simulations. Herein, two distinct approaches—the penalty and dissipation approach—are applied to the thrombus growth, with a comparison made to determine the most suitable method. The penalty approach is based on a fibrin-linked velocity penalty term while in the dissipation approach the Einstein equation is linked with fibrin concentration. The model simulates the coagulation cascade by accounting for concentrations of key elements such as thrombin, prothrombin, fibrinogen, fibrin, and both activated and resting platelets. The implementation is carried out using the open-source DualSPHysics solver, incorporating the wall shear stress effects alongside thrombus development. To validate the model, simulations of thrombus formation were conducted in a backward-facing step and a microchannel. The results demonstrate the potential of SPH and the proposed approach in transforming thrombus modeling, particularly for predicting device-induced thrombosis. This research highlights its potential to advance the understanding of cardiovascular diseases and improve clinical outcomes.

A review of emission characteristics and risk assessments of volatile organic compounds in petrochemical industry areas.

As the petrochemical industry grows, environmental and human health issues associated with petroleum refining and chemical processes also increase. Consequently, several studies have been conducted on this topic. However, the results of the current research vary, and a comprehensive review is lacking. This study summarized the volatile organic compounds (VOCs) emission and risk assessments in the petrochemical industry based on data collected from previous studies. A discussion of VOC emission characteristics is provided. The effects of VOCs on human health, ozone formation potential (OFP), and secondary organic aerosol (SOA) are also reviewed. According to this review, the VOC emission characteristics are related to the raw materials and processes. Moreover, research methods can lead to certain biases. In entire petrochemical plants, alkanes were the largest contributors to VOC emissions, with n-pentane, n-butane, and propane frequently appearing in the top five emission lists. Among the process unit areas, alkanes were the major contributors, except for the delayed coking unit, where aromatics significantly contributed. Regarding the risks associated with VOC emissions, benzene, and 1,3-butadiene are common carcinogens impacting human health. 1,3-Butadiene, benzene, and acrolein are major contributors to noncarcinogenic risk. OFP is related to VOC emissions and their corresponding reactivities. Alkenes, alkanes, and aromatics are major contributors to OFP. Aromatics were the largest contributors to SOA concentration. In the future, research methods on the characteristics and risks of VOC emissions need to be further improved. More precise sampling techniques and advanced analytical instruments should be employed to better characterize VOC emissions. Overall, this study considered the characteristics of VOC emissions from petrochemical industrial areas, as well as the risks to the environment and health to provide a reference for the control of VOCs.

Super-resolution left ventricular flow and pressure mapping by Navier-Stokes-informed neural networks.

Intraventricular vector flow mapping (VFM) is an increasingly adopted echocardiographic technique that derives time-resolved two-dimensional flow maps in the left ventricle (LV) from color-Doppler sequences. Current VFM models rely on kinematic constraints arising from planar flow incompressibility. However, these models are not informed by crucial information about flow physics; most notably the forces within the fluid and the resulting accelerations. This limitation has rendered VFM unable to combine information from different time frames in an acquisition sequence or derive fluctuating pressure maps. In this study, we leveraged recent advances in artificial intelligence (AI) to develop AI-VFM, a vector flow mapping modality that uses physics-informed neural networks (PINNs) encoding mass conservation and momentum balance inside the LV, and no-slip boundary conditions at the LV endocardium. AI-VFM recovers the flow and pressure fields in the LV from standard echocardiographic scans. It performs phase unwrapping and recovers flow data in areas without input color-Doppler data. AI-VFM also recovers complete flow maps at time points without color-Doppler input data, producing super-resolution flow maps. We validate AI-VFM using physiological simulated LV data and show that informing the PINNs with momentum balance is essential for achieving temporal super-resolution and significantly increases the accuracy of AI-VFM compared to informing the PINNs only with mass conservation. AI-VFM is solely informed by each patient's flow physics; it does not utilize explicit smoothness constraints or incorporate data from other patients or flow models. AI-VFM takes 15 min to run in off-the-shelf graphics processing units and its underlying PINN framework could be extended to map other flow-associated metrics such as blood residence time or the concentration of coagulation species.

A multi-scale feature fusion network based on semi-channel attention for seismic phase picking

In the field of seismic data processing, deep learning technologies have been widely used for seismic phase picking. However, it is difficult to take full advantage of the features extracted at different stages in existing models. In this paper, a multi-scale feature fusion network was proposed for seismic phase picking to address this problem. In the stage of feature extraction, semi-channel attention is introduced. It improves the representation ability of the model by efficiently utilizing the feature information extracted from the encoder. In the stage of decoding, a channel compression module is designed to reduce the number of feature channels. It improves the receptive field of channels. Additionally, a multi-feature fusion module is presented to integrate features at multiple scales. It reduces the loss of useful information and improves the accuracy of phase picking. The effectiveness of our network is validated on Stanford earthquake dataset, where the picking errors for phase picking are 2 ms. The parameter of our network is only 52,100. Compared with earthquake transformer, it has 42.1% fewer time costs to process 12,656 test samples on Graphics Processing Unit.

Applications of multi-GPU computing in large-scale fully kinetic particle-in-cell simulations of space plasma

The rapid development of an emerging computing device, the graphical processing unit (GPU), has significantly enhanced our ability to conduct full kinetic particle-in-cell (PIC) simulations in space physics. In this paper, we propose an approach that leverages multiple GPUs to facilitate large-scale PIC simulations. This method can effectively reduce data transmission frequency and latency during the computing process. The data communication between GPU devices is optimized through a combination of Message Passing Interface (MPI)-NVIDIA Collective Communications Library (NCCL) running pattern. Our implementation surpasses the expected linear acceleration, achieving superior computing performance and operational efficiency. The instances of large-scale PIC simulations are presented based on physical models of magnetic reconnection, plasma turbulence, and quasi-perpendicular shock. The importance of large-scale simulations is demonstrated in terms of grid resolution, macroparticles used per cell, and the mass ratio between ions and electrons. The multi-GPU enabled fully kinetic PIC simulation demonstrates its capability to efficiently handle large-scale PIC simulations as a crucial requirement for the study of space plasma physics.

Parallel rapidly exploring random tree method for unmanned aerial vehicles autopilot development using graphics processing unit processing

Parallel rapidly exploring random tree method for unmanned aerial vehicles autopilot development using graphics processing unit processing

An Efficient Large Kernel Convolution Network Designed for Neural Processing Unit

An Efficient Large Kernel Convolution Network Designed for Neural Processing Unit