Computing On Graphics Processing Units Research Articles

Efficient computation of the geopotential gradient in graphic processing units

Efficient computation of the geopotential gradient is essential for numerical propagators, particularly in scenarios involving low Earth orbits. Conventional geopotential calculations are based on spherical harmonics series, which become computationally demanding as the degree/order increases. This computational burden can be mitigated by means of parallelized algorithms. Additionally, certain situations lend themselves to high parallelization, such as the propagation of space debris catalogs, satellite mega-constellations, or the dispersion of particles resulting from a space collision event. This paper introduces an optimized Graphics Processing Unit (GPU) implementation designed to facilitate extensive parallelization in the geopotential gradient calculation. The formulation developed in this study is not specific to any GPU. However, to illustrate the low-level optimizations necessary for an efficient implementation, we selected the Compute Unified Device Architecture (CUDA) as the dominant and de facto standard in parallel computing. Nevertheless, most of the concepts and optimizations presented in this paper are also valid for other GPU architectures. Built upon the spherical harmonic expansion using the Cunningham formulation, which is well-suited for GPU computations, our implementation offers several variants with different tradeoffs between speed and accuracy. Besides GPU double precision, we introduced a mixed precision arithmetic –a hybrid between single and double precision– that exploits GPU capabilities with a low penalty in accuracy. The proposed algorithm was implemented as a software reusable module, and its performance was evaluated against GMAT, GODOT, and Orekit astrodynamic codes. The algorithm’s accuracy in double precision is comparable to such codes. The mixed precision version showed enough accuracy for LEO satellite propagation, with around 1 m difference in four days. Testing across different CUDA architectures revealed very high speed-up factors compared to a single CPU, reaching a speed-up of 645 for the mixed precision variant and 450 for the double precision one in the propagation of about 3200 objects with a geopotential of degree/order 126×126 using an A100 GPU device.

Uncertainty Analysis of Post-Failure Behavior in Landslides Based on SPH Method and Generalized Geotechnical Random Field Theory

In slope stability analysis, the inherent heterogeneity and spatial variability of soil significantly influence the aftermath of landslides, including critical aspects like runout distance, influence distance, and volume. This research integrates Smoothed Particle Hydrodynamics (SPH) with random field theory to precisely model the large deformation events in slopes with anisotropic shear parameters, while leveraging Graphics Processing Unit (GPU) parallel computing to expedite the generation of random field samples and SPH simulations. This approach meticulously examines how spatial heterogeneity of soil affects slope instability, simulating soil parameters across varied geological settings. It considers the fluctuation scale of anisotropy, the cross-correlation of cohesion and the angle of internal friction, along with their coefficient of variation (CV), to elucidate their impact on landslide magnitude and severity. This study furnishes a robust tool for a holistic assessment of slope impacts and landslide volumes. Additionally, by delineating the computational efficiency disparities between GPUs and Central processing units (CPUs), it underscores the pronounced efficiency benefits of GPU computing. The insights garnered offer fresh perspectives and methodologies in slope stability analysis and disaster risk evaluation, contributing to the finer prediction and management of this prevalent and grave geological hazard.

Advanced Framework for Multi-Modal Healthcare Data Integration: Leveraging HPC with GPU Computing and CNN Architecture in CDSS

In this study, we shall be looking at the challenges involved in integrating multi-modal healthcare data in the clinical decision support systems (CDSS). We propose the Automated Multi-Modal Data Integration (AMMI-CDSS) algorithm, which will utilize the latest high-performance computing (HPC) techniques such as the Convolutional Neural Network (CNN) architecture and the Graphics Processing Unit (GPU) computing to provide precise and rapid analysis. Which features will be extracted, multi-modal data will be merged, data will be prepared and algorithms developed in a distributed computing environment. We illustrate how AMMI-CDSS through the use of real world datasets such as wearable sensors data, medical imaging, genetic data, and electronic health records (EHRs), can improve the clinical decision support. By performing harmonization of the diverse data sources into a unique dataset after thorough data preprocessing and complex calculations, AMMI-CDSS provides the analysis with better quality and coherence.
 Our study allow us to make conclusion about how HPC-based CDSS models can be compared to conventional machine learning ones using their scalability and performance as key metrics. We enrich CDSS with the methodical framework for one-by-one testing and evaluation of proposed models and multi-modal healthcare data analysis. Future research might explore novel methods for integrating diverse types of healthcare data, as well as enhancing the HPC-based CDSS models by keeping them up-to-date.

A GPU-accelerated computational fluid dynamics solver for assessing shear-driven indoor airflow and virus transmission by scale-resolved simulations

We explore the applicability of MATLAB for 3D computational fluid dynamics (CFD) of shear-driven indoor airflows. A new scale-resolving, large-eddy simulation (LES) solver titled DNSLABIB is proposed for MATLAB utilizing graphics processing units (GPUs). In DNSLABIB, the finite difference method is applied for the convection and diffusion terms while a Poisson equation solver based on the fast Fourier transform (FFT) is employed for the pressure. The immersed boundary method (IBM) for Cartesian grids is proposed to model solid walls and objects, doorways, and air ducts by binary masking of the solid/fluid domains. The solver is validated in two canonical reference cases and against experimental data. Then, we demonstrate the validity of DNSLABIB in a room geometry by comparing the results against another CFD software (OpenFOAM). Next, we demonstrate the solver performance in several isothermal indoor ventilation configurations and the implications of the results are discussed in the context of airborne transmission of COVID-19. The novel numerical findings using the new CFD solver are as follows. First, a linear scaling of DNSLABIB is demonstrated and a speed-up by a factor of 3-4 is also noted in comparison to similar OpenFOAM simulations. Second, ventilation in three different indoor geometries are studied at both low (0.1 m/s) and high (1 m/s) airflow rates corresponding to Re=5000 and Re=50000. An analysis of the indoor CO2 concentration is carried out as the room is emptied from stale, high CO2 content air. We estimate the air changes per hour (ACH) values for three different room geometries and show that the numerical estimates from 3D CFD simulations differ by 80%–150% (Re=50000) and 75%–140% (Re=5000) from the theoretical ACH value based on the perfect mixing assumption. Third, the analysis of the CO2 probability distributions (PDFs) indicates a relatively non-uniform distribution of fresh air indoors. Fourth, utilizing a time-dependent Wells-Riley analysis, an example is provided on the growth of the cumulative infection risk which is shown to reduce rapidly after the ventilation is started. The average infection risk is shown to reduce by a factor of 2 for lower ventilation rates (ACH=3.4-6.3) and 10 for the higher ventilation rates (ACH=37-64). Finally, we utilize the new solver to comment on respiratory particle transport indoors. A key contribution of the paper is to provide an efficient, GPU compatible CFD solver environment enabling scale-resolved simulations (LES/DNS) of airflow in large indoor geometries on a single GPU designed for high-performance computing. The demonstrated efficacy of MATLAB for GPU computing indicates a high potential of DNSLABIB for various future developments on airflow prediction.

Faster inference from state space models via GPU computing

Inexpensive Graphics Processing Units (GPUs) offer the potential to greatly speed up computation by employing their massively parallel architecture to perform arithmetic operations more efficiently. Population dynamics models are important tools in ecology and conservation. Modern Bayesian approaches allow biologically realistic models to be constructed and fitted to multiple data sources in an integrated modelling framework based on a class of statistical models called state space models. However, model fitting is often slow, requiring hours to weeks of computation. We demonstrate the benefits of GPU computing using a model for the population dynamics of British grey seals, fitted with a particle Markov chain Monte Carlo algorithm. Speed-ups of two orders of magnitude were obtained for estimations of the log-likelihood, compared to a traditional ‘CPU-only’ implementation, allowing for an accurate method of inference to be used where this was previously too computationally expensive to be viable. GPU computing has enormous potential, but one barrier to further adoption is a steep learning curve, due to GPUs' unique hardware architecture. We provide a detailed description of hardware and software setup, and our case study provides a template for other similar applications. We also provide a detailed tutorial-style description of GPU hardware architectures, and examples of important GPU-specific programming practices.

A GPU scheme for multi-secret visual sharing with varied secret dimensions and contrast enhancement using blind super-resolution

Multi-secret visual sharing schemes are essential for secure and efficient sharing of sensitive visual information. However, existing schemes often overlook important considerations such as varied secret dimensions, contrast enhancement, and computational efficiency. This research addresses these challenges by proposing a comprehensive approach that addresses these limitations. Firstly, the scheme takes into account the varied dimensions of secret images encountered in real-world scenarios, allowing flexibility in sharing and reconstructing images of different sizes and aspect ratios. Secondly, the research integrates contrast enhancement techniques, such as blind super-resolution, to improve the visual quality and visibility of shared secret images affected by factors like noise, compression, or low lighting conditions. Lastly, to enhance the computational efficiency, the scheme leverages the power of Graphics Processing Units (GPUs) for parallel computing, enabling faster processing of large-scale image operations. The proposed scheme achieves outstanding results, with a high PSNR of 98.264, a strong NCC value of 0.965, an exceptionally low NAE value of 0.046, and an impressive SSIM value of 0.983. Furthermore, the GPU implementation provides a remarkable overall speedup of 112×\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$112\ imes$$\\end{document} for 256×256\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$256 \ imes 256$$\\end{document} color images.

Parallel computation to bidimensional heat equation using MPI/CUDA and FFTW package

In this study, we present a fast algorithm for the numerical solution of the heat equation. The heat equation models the heat diffusion over time and through a given region. We engage a finite difference method to solve this equation numerically. The performance of its parallel implementation is considered using Message Passing Interface (MPI), Compute Unified Device Architecture (CUDA), and time schemes, such as Forward Euler (FE) and Runge-Kutta (RK) methods. The originality of this study is research on parallel implementations of the fourth-order Runge-Kutta method (RK4) for sparse matrices on Graphics Processing Unit (GPU) architecture. The supreme proprietary framework for GPU computing is CUDA, provided by NVIDIA. We will show three metrics through this parallelization to compare the computing performance: time-to-solution, speed-up, and performance. The spectral method is investigated by utilizing the FFTW software library, based on the computation of the fast Fourier transforms (FFT) in parallel and distributed memory architectures. Our CUDA-based FFT, named CUFFT, is performed in platforms, which is a highly optimized FFTW implementation. We will give numerical tests to reveal that this method is up-and-coming for solving the heat equation. The final result demonstrates that CUDA has a significant advantage and performance since the computational cost is tiny compared with the MPI implementation. This vital performance gain is also achieved through careful attention of managing memory communication and access.

Accelerating Static Timing Analysis Using CPU–GPU Heterogeneous Parallelism

Static timing analysis (STA) is an essential yet time-consuming task during the circuit design flow to ensure the correctness and performance of the design. Thanks to the advancement of general-purpose computing on graphics processing units (GPUs), new possibilities and challenges have arisen for boosting the performance of STA. In this work, we present an efficient and holistic GPU-accelerated STA engine. We accelerate major STA tasks, including levelization, delay computation, graph propagation, and multi-corner analysis, by developing high-performance GPU kernels and data structures. By dividing the STA workloads into CPU-GPU concurrent tasks with managed dependencies, our acceleration framework supports versatile incremental updates. Furthermore, we have extended our approach to multi-corner analysis by exploring a large amount of corner-level data parallelism using GPU computing. Our implementation based on the open-source STA engine OpenTimer has achieved up to 4.07× speed-up on single corner analysis, and up to 25.67× speed-up on multi-corner analysis on TAU 2015 contest designs and a 14nm technology.

GPIC: A set of high-efficiency CUDA Fortran code using gpu for particle-in-cell simulation in space physics

Previous implementation of the full kinetic particle-in-cell (PIC) simulation on the Graphical Processing Unit (GPU) device has shown advantages compared to traditional Central Processing Unit (CPU) computing. However, conventional PIC simulations using GPU computing have faced limitations in terms of low performance when the simulation box size or the number of particles per cell increases. In this study, we boost computing efficiency by designing novel schemes of kernels with the combination of numerical and technical aspects. The time expenses of the global data transferring and duplicate data fetching processes have been massively reduced by the utilization of the on-chip memory. The reduction treatment on the particles, the graded current computing scheme, and the 2.5D thread launch strategy are designed to accelerate the simulation iterations. The new scheme can reach up to about 5.5 times the acceleration ratio than the old one and attain the highest to 734 times faster than the program using CPUs. Our new scheme can realize large-scale PIC simulations with both decent performances on the legacy GPUs and the increasing trend of the acceleration ratio on the lasted GPUs.

Development of scalable GPU-based direct whole-core depletion calculation methods

GPU (Graphical Processing Unit) acceleration methods are introduced to the depletion solver and the cross section (XS) calculation routines of the direct whole core calculation code, nTRACER, for scalable and superior computing performance. Extensive GPU acceleration besides transport solvers is necessary to resolve the performance degradation in depletion calculations caused by the newly generated nuclides in depleted fuel regions and the poor scalability. A matrix exponential solver employing the Gauss-Seidel-based Chebyshev rational approximate method is implemented to reduce branch divergence and load imbalance in GPU computing in the solution of the Bateman equation. A non-zero element major storage scheme is developed to optimize memory access efficiency on GPUs by exploiting the same sparsity pattern of the depletion matrix in all the depletion regions. XS data are serialized for effective massive parallelization of XS calculation on GPUs. A block decomposition scheme for XS calculation is implemented to reduce the global memory requirement due to the limited memory capacity of commercial GPUs. It is demonstrated that nTRACER can complete a direct whole core depletion calculation for a realistic core in a few hours on a moderate-sized GPU-based heterogeneous computing platform after extensive GPU porting to all the essential routines.

MRT-LB simulation and response surface analysis of natural convection of non-Newtonian ferrofluid in an enclosure with non-uniformly heated radiator through GPU computing

The multiple-relaxation-time (MRT) lattice Boltzmann method (LBM) has been applied to numerically analyze the natural convection of non-Newtonian ferrofluid in an enclosure with a horizontal plate or radiator that is not uniformly heated. The left and right walls of the enclosure are cold (Tc), while the bottom and upper walls are assumed to be adiabatic. Simulations using numerical models are run for a variety of dimensionless variables, including Rayleigh numbers (Ra=104,105,106), Hartmann numbers (Ha=0,20,40,60,80), power-law index (n=0.6,0.8,1.0,1.2,1.4), magnetic field inclination angles (γ=0,π/4,π/2,3π/4), and the maximum ferroparticle volume fractions is 6% (ϕ=0.06). The numerical results are reported in the form of isotherms, streamlines, velocity, temperature, mean Nusselt number, and entropy production. The study found that the Nu¯ rises with the increment of the Rayleigh number but decreases with the enhancement of the Ha, n, and ϕ. The temperature declines rapidly when the magnetic field inclination angle values increase to γ=90°, then gains temperature. The velocity and temperature both increase as the ϕ increases. For increasing the power-law index, the Nu¯ grows by 1.6% in the case of Ra=104 but falls by 14.56% and 54.87% when Ra=105 and Ra=106, respectively. The entropy profile is discussed in detail. The Be number is more significant than 0.5 for various n at Ra=104 and 105, suggesting that entropy formation is primarily due to heat gradients. Response surface methodology (RSM) also describes the numerical results with variable Ra numbers (104, 105, 106), n=0.6,1.0,1.4), angles (γ=0,π/2,3π/4), Ha=40 and ϕ=0.06. Finally, the residual plots, 2D and 3D plots are described. All the simulations are conducted by the GPU (Graphics Processing Unit) computing with the NVIDIA CUDA C/C++ platform.

A hybrid CPU/GPU method for Hartree-Fock self-consistent-field calculation.

The calculation of two-electron repulsion integrals (ERIs) is a crucial aspect of Hartree-Fock calculations. In computing the ERIs of varying angular momentum, both the central processing unit (CPU) and the graphics processing unit (GPU) have their respective advantages. To accelerate the ERI evaluation and Fock matrix generation, a hybrid CPU/GPU method has been proposed to maximize the computational power of both CPU and GPU while overlapping the CPU and GPU computations. This method employs a task queue where each task corresponds to ERIs with the same angular momentum. The queue begins with ERIs of low angular momentum, which are computationally efficient on GPUs, and ends with ERIs of high angular momentum, which are better suited for CPU computation. CPUs and GPUs dynamically grab and complete tasks from the start and end of the queue using OpenMP dynamic scheduling until all tasks are finished. The hybrid CPU/GPU computation offers the advantage of enabling calculations with arbitrary angular momentum. Test calculations showed that the hybrid CPU/GPU algorithm is more efficient than "GPU-only" when using a single GPU. However, as more GPUs are involved, the advantage diminishes or disappears. The scaling exponents of the hybrid method were slightly higher than "GPU-only," but the pre-exponent factor was significantly lower, making the hybrid method more effective overall.

Adopting GPU computing to support DL-based Earth science applications

ABSTRACT With the advancement of Artificial Intelligence (AI) technologies and accumulation of big Earth data, Deep Learning (DL) has become an important method to discover patterns and understand Earth science processes in the past several years. While successful in many Earth science areas, AI/DL applications are often challenging for computing devices. In recent years, Graphics Processing Unit (GPU) devices have been leveraged to speed up AI/DL applications, yet computational performance still poses a major barrier for DL-based Earth science applications. To address these computational challenges, we selected five existing sample Earth science AI applications, revised the DL-based models/algorithms, and tested the performance of multiple GPU computing platforms to support the applications. Application software packages, performance comparisons across different platforms, along with other results, are summarized. This article can help understand how various AI/ML Earth science applications can be supported by GPU computing and help researchers in the Earth science domain better adopt GPU computing (such as supermicro, GPU clusters, and cloud computing-based) for their AI/ML applications, and to optimize their science applications to better leverage the computing device.

Direct numerical simulation of slender cones with variable nose bluntness based on graphics processing unit computation

Direct numerical simulation with up to 10×109 scale grid points based on graphics processing unit computation is carried out to investigate the bluntness effect on the hypersonic boundary-layer transition over a slender cone with zero angle of attack at Mach 6. Four cases with the nose radii of 1, 10, 20, and 40 mm are conducted, and the corresponding Reynolds number based on the nose radius varies from 1.0×104 to 4.0×105. Random disturbances with a broad spectrum of frequencies and a wide range of azimuthal wavenumbers were applied to the wall to simulate disturbances caused by wall roughness. The numerical results show that as the nose tip radius increases, the transition position gradually moves downstream with increased transition region. For the case with a nose radius of 1 mm, the flow transition and entropy swallowing occur almost simultaneously, while for other cases, the transition takes place earlier than the entropy swallowing. In consequence, the disturbance amplitude upstream of the transition in the 1 mm case is much larger than that of other cases. To further study the mechanism of the transition, the frequency spectrum analysis is carried out. It is found that all cases exhibit two characteristic frequencies within the transition region, i.e., the high frequency and extremely low frequency. Owing to the influence of the entropy layer, the characteristic high frequency of the 1 mm case is significantly higher than that of other cases. With the increase in the nose radius, the characteristic frequency of the high frequency decreases gradually.

Kohn–Sham time-dependent density functional theory with Tamm–Dancoff approximation on massively parallel GPUs

We report a high-performance multi graphics processing unit (GPU) implementation of the Kohn–Sham time-dependent density functional theory (TDDFT) within the Tamm–Dancoff approximation. Our algorithm on massively parallel computing systems using multiple parallel models in tandem scales optimally with material size, considerably reducing the computational wall time. A benchmark TDDFT study was performed on a green fluorescent protein complex composed of 4353 atoms with 40,518 atomic orbitals represented by Gaussian-type functions, demonstrating the effect of distant protein residues on the excitation. As the largest molecule attempted to date to the best of our knowledge, the proposed strategy demonstrated reasonably high efficiencies up to 256 GPUs on a custom-built state-of-the-art GPU computing system with Nvidia A100 GPUs. We believe that our GPU-oriented algorithms, which empower first-principles simulation for very large-scale applications, may render deeper understanding of the molecular basis of material behaviors, eventually revealing new possibilities for breakthrough designs on new material systems.

Solving the Neutron Transport Equation for Microreactor Modeling Using Unstructured Meshes and Exascale Computing Architectures

The Microreactor Exascale eZ CALculation (MEZCAL) tool has been developed to accurately and efficiently solve the neutron transport equation in general, unstructured meshes to support the design and modeling of microreactors. MEZCAL solves the self-adjoint angular flux form of the neutron transport equation using the finite element method. As the neutron transport equation is computationally expensive to solve, MEZCAL is designed to efficiently use exascale computing architectures, with an emphasis on graphics processing unit computing. To leverage existing tools, MEZCAL is built using the MFEM library and uses solvers from HYPRE, PETSc, and SLEPc. Verification of the neutron transport solver in MEZCAL is demonstrated with the solution to a one-dimensional cylindrical problem that has a semi-analytic solution. After verification, a realistic microreactor based on the MARVEL microreactor design is modeled using MEZCAL. Spatial and angular refinement results are presented for a two-dimensional model of the MARVEL microreactor, and the eigenvalue is converged to approximately 60 pcm. This convergence required a very fine mesh and more than 3.76 Billion Degrees Of Freedom (BDOF). Preliminary results are also presented for a three-dimensional model of the MARVEL microreactor. Finally, a weak scaling study is performed to investigate how the methods in MEZCAL will scale for larger problems with the next generation of exascale computing architectures.

Graph Traversal (GT) on Linear and Quadratic Graphic Processing Units with CPU

Nowadays, Graph Traversal (GT) is a fundamental procedure for developing a measure in computing systems and algorithms. Graph traversal algorithms usually iterate input datasets of a graph to provide a logical convergence in every iteration. Graphic Processing Units (GPUs) have been utilized broadly as accelerators of GT due to their massive parallelism and accessibility of high-transfer speed memory. Furthermore, graph-based computation is pervasive and challenging for memory systems implementation in a specific time period. This paper provides quadratic and single GPUs, allowing developers to extend graph-based GPU computation effortlessly. The paper concentrates on the creditability and validation of sparse graphs in GPUs. Since the benchmark characteristic of Breadth-First Search (BFS) plays a critical role in graphs, this paper considers a primary characteristic benchmark for BFS computation usage. This paper explores the effectiveness and applicability of BFS for analyzing graph systems by utilizing the mentioned characteristics in BFS computation. This paper also utilizes the penalization of GPU BFS during the analysis and simulation results. In addition, using a comparative simulation result, this paper concludes that improved traversal rates speed up the computation and reduce memory usage for the single and quadratic GPU structures.

The W transform with a chirp-modulated window

A high-resolution time–frequency spectrum is desirable for processing and interpreting seismic data. The W transform is a method that effectively preserves the resolution in the low-frequency region of the time–frequency spectrum of nonstationary seismic signals. To further increase the energy concentration of the time–frequency spectrum estimated with the W transform, we propose to combine the W transform with a chirp-modulated window. The chirp rate in the chirp-modulated window can control the rotation of the window in the time–frequency plane to achieve a better match with the time–frequency spectrum. The chirp rate in the modified W transform can be directly determined with the estimated instantaneous frequency. It has been shown that the W transform with a chirp-modulated window maintains the resolution of the time–frequency spectrum and improves the energy concentration around the dominant frequency against noise. To speed up the computational process of the W transform with a chirp-modulated window, we formulate the transform as a matrix-vector multiplication, which can be accelerated by using graphics processing unit computations. The application of the modified algorithm to synthetic and field data indicates that the frequency anomalies can be easily identified with the modified W transform.

B2-Net: an artificial intelligence powered machine learning framework for the classification of pneumonia in chest x-ray images

A chest x-ray radiograph is still the global standard for diagnosing pneumonia and helps distinguish between bacterial and viral pneumonia. Despite several studies, radiologists and physicians still have trouble correctly diagnosing and classifying pneumonia without false negatives. Modern mathematical modeling and artificial intelligence could help to reduce false-negative rates and improve diagnostic accuracy. This research aims to create a novel and efficient multiclass machine learning framework for analyzing and classifying chest x-ray images on a graphics processing unit (GPU). Researchers initially applied a geometric augmentation using a positional transformation function to the original dataset to enhance the sample size and aid future transfer learning. Models with the best accuracy, area under the receiver operating characteristics (AUROC), F1 score, precision, recall, and specificity are chosen from a pool of nine state-of-the-art neural network models. The best-performing models are then retrained using an ensemble technique using depth-wise convolutions, demonstrating significant improvements over the baseline models employed in this research. With a remarkable 97.69% accuracy, 100% recall, and 0.9977 AUROC scores, the proposed Bek-Bas network (B2-Net) model can differentiate between normal, bacterial, and viral pneumonia in chest x-ray images. A superior model is retrained using the chosen dense convolutional network-160, residual network-121, and visual geometry group network-16 ensemble models. The diagnostic accuracy of the x-ray classification unit is enhanced by the newly designed multiclass network, the B2-Net model. The developed GPU-based framework has been examined and tested to the highest clinical standards. After extensive clinical testing, the final B2-Net model is implemented on an NVIDIA Jetson Nano GPU computer. Healthcare facilities have confirmed the B2-Net is the most effective framework for identifying bacterial and viral pneumonia in chest x-rays.

A Hybrid GPU and CPU Parallel Computing Method to Accelerate Millimeter-Wave Imaging

The range migration algorithm (RMA) based on Fourier transformation is widely applied in millimeter-wave (MMW) close-range imaging because of its few operations and small approximation. However, its interpolation stage is not effective due to the involved intensive logic controls, which limits the speed performance in a graphics processing unit (GPU) platform. Therefore, in this paper, we present an acceleration optimization method based on the hybrid GPU and central processing unit (CPU) parallel computation for implementing the RMA. The proposed method exploits the strong logic-control capability of the CPU to assist the GPU in processing the logic controls of the interpolation stage. The common positions of wavenumber-domain components to be interpolated are calculated by the CPU and stored in the constant memory for broadcast at any time. This avoids the repetitive computation consumed in a GPU-only scheme. Then the GPU is responsible for the remaining matrix-related steps and outputs the needed wavenumber-domain values. The imaging experiments verify the acceleration efficiency of the proposed method and demonstrate that the speedup ratio of our proposed method is more than 15 times of that by the CPU-only method, and more than 2 times of that by the GPU-only method.