Search Results (244)

The Winograd algorithm reduces the computational complexity of convolutional neural networks (CNNs) by minimizing the number of multiplications required for convolutions, making it particularly suitable for resource-constrained edge devices. Concurrently, most edge hardware accelerators utilize 8-bit integer arithmetic to enhance energy efficiency and reduce inference latency, requiring the quantization of CNNs before deployment. Combining Winograd-based convolution with quantization offers the potential for both performance acceleration and reduced energy consumption. However, prior research has identified significant challenges in this combination, particularly due to numerical instability and substantial accuracy degradation caused by the transformations required in the Winograd domain, making the two techniques incompatible on edge hardware. In this work, we describe our latest training scheme, which addresses these challenges, enabling the successful integration of Winograd-accelerated convolution with low-precision quantization while maintaining high task-related accuracy. Our approach mitigates the numerical instability typically introduced during the transformation, ensuring compatibility between the two techniques. Additionally, we extend our work by presenting a custom-optimized CUDA implementation of quantized Winograd convolution for NVIDIA edge GPUs. This implementation takes full advantage of the proposed training scheme, achieving both high computational efficiency and accuracy, making it a compelling solution for edge-based AI applications. Our training approach enables significant MAC reduction with minimal impact on prediction quality. Furthermore, our hardware results demonstrate up to a 3.4× latency reduction for specific layers, and a 1.44× overall reduction in latency for the entire DeepLabV3 model, compared to the standard implementation. Full article

In this paper, we consider the finite difference approximation for a one-dimensional mathematical model of heat conduction in a three-layered solid with interfacial conditions for temperature and heat flux between the layers. The finite difference scheme is unconditionally stable, convergent, and equivalent to the solution of two linear algebraic systems. We evaluate various methods for solving the involved linear systems by analyzing direct and iterative solvers, including GPU-accelerated approaches using CuPy and PyCUDA. We evaluate performance and scalability and contribute to advancing computational techniques for modeling complex physical processes accurately and efficiently. Full article

We introduce <tt>ssc-cdi</tt>, an open-source software package from the Sirius Scientific Computing family, designed for memory-efficient, single-node multi-GPU ptychography reconstruction. <tt>ssc-cdi</tt> offers a range of reconstruction engines in Python version 3.9.2 and C++/CUDA. It aims at developing local expertise and customized solutions to meet the specific needs of beamlines and user community of the Brazilian Synchrotron Light Laboratory (LNLS). We demonstrate ptychographic reconstruction of beamline data and present benchmarks for the package. Results show that <tt>ssc-cdi</tt> effectively handles extreme datasets typical of modern X-ray facilities without significantly compromising performance, offering a complementary approach to well-established packages of the community and serving as a robust tool for high-resolution imaging applications. Full article

Efficiently simulating solid mechanics is vital across various engineering applications. As constitutive models grow more complex and simulations scale up in size, harnessing the capabilities of modern computer architectures has become essential for achieving timely results. This paper presents advancements in running parallel simulations of solid mechanics on multi-core CPUs and GPUs using a single-code implementation. This portability is made possible by the C++ matrix and array (MATAR) library, which interfaces with the C++ Kokkos library, enabling the selection of fine-grained parallelism backends (e.g., CUDA, HIP, OpenMP, pthreads, etc.) at compile time. MATAR simplifies the transition from Fortran to C++ and Kokkos, making it easier to modernize legacy solid mechanics codes. We applied this approach to modernize a suite of constitutive models and to demonstrate substantial performance improvements across different computer architectures. This paper includes comparative performance studies using multi-core CPUs along with AMD and NVIDIA GPUs. Results are presented using a hypoelastic–plastic model, a crystal plasticity model, and the viscoplastic self-consistent generalized material model (VPSC-GMM). The results underscore the potential of using the MATAR library and modern computer architectures to accelerate solid mechanics simulations. Full article

Background/Objectives: The current research examines the accuracy of α-synuclein in RBCs as a diagnostic biomarker for PD and PSP, despite their distinct molecular etiologies. Methods: We used ELISA to measure total, oligomeric, and p129-α-synuclein levels in erythrocytes from 8 PSP patients, 19 PD patients, and 18 healthy controls (HCs). The classification performances of RBC α-synuclein levels were investigated by receiver operator characteristic (ROC) curve. We also evaluated a possible correlation between RBC α-synuclein level and the biological and clinical features of our cohorts. Results: RBC total α-synuclein was higher in PSP patients compared to both PD patients and HCs, achieving good classification performance (AUC: 0.853) in distinguishing PSP patients from PD patients, with a sensitivity of 100% and a specificity of 70.6%; moreover, the levels of this biomarker positively correlated with disease severity in PSP group. Regarding oligomeric α-synuclein and p129-α-synuclein, the latter was slightly increased in RBCs from PSP patients compared to HCs, but no correlations were detected. Conclusions: Although these findings need to be confirmed in larger studies, our pilot work suggests that RBC total α-synuclein may represent a potential molecular biomarker for the differential diagnosis and clinical staging of PSP. Full article

Theano, TensorFlow, Keras, Torch, PyTorch, and other software frameworks have remarkably stimulated the popularity of deep learning (DL). Apart from all the good they achieve, the danger of such frameworks is that they unintentionally spur a black-box attitude. Some practitioners play around with building blocks offered by frameworks and rely on them, having a superficial understanding of the internal mechanics. This paper constitutes a concise tutorial that elucidates the flows of signals and gradients in deep neural networks, enabling readers to successfully implement a deep network from scratch. By “from scratch”, we mean with access to a programming language and numerical libraries but without any components that hide DL computations underneath. To achieve this goal, the following five topics need to be well understood: (1) automatic differentiation, (2) the initialization of weights, (3) learning algorithms, (4) regularization, and (5) the organization of computations. We cover all of these topics in the paper. From a tutorial perspective, the key contributions include the following: (a) proposition of R and S operators for tensors—rashape and stack, respectively—that facilitate algebraic notation of computations involved in convolutional, pooling, and flattening layers; (b) a Python project named hmdl (“home-made deep learning”); and (c) consistent notation across all mathematical contexts involved. The hmdl project serves as a practical example of implementation and a reference. It was built using NumPy and Numba modules with JIT and CUDA amenities applied. In the experimental section, we compare hmdl implementation to Keras (backed with TensorFlow). Finally, we point out the consistency of the two in terms of convergence and accuracy, and we observe the superiority of the latter in terms of efficiency. Full article

Clinical differentiation of progressive supranuclear palsy (PSP) from Parkinson’s disease (PD) is challenging due to overlapping phenotypes and late onset of PSP specific symptoms, highlighting the need for easily assessable biomarkers. We used proximity elongation assay (PEA) to analyze 460 proteins in serum samples from 46 PD, 30 PSP patients, and 24 healthy controls. ANCOVA was used to identify the most promising proteins and machine learning (ML) XGBoost and random forest algorithms to assess their classification performance. Promising proteins were also quantified by ELISA. Moreover, correlations between serum biomarkers and biological and clinical features were investigated. We identified five proteins (TFF3, CPB1, OPG, CNTN1, TIMP4) showing different levels between PSP and PD, which achieved good performance (AUC: 0.892) when combined by ML. On the other hand, when the three most significant biomarkers (TFF3, CPB1 and OPG) were analyzed by ELISA, there was no difference between groups. Serum levels of TFF3 positively correlated with age in all subjects’ groups, while for OPG and CPB1 such a correlation occurred in PSP patients only. Moreover, CPB1 positively correlated with disease severity in PD, while no correlations were observed in the PSP group. Overall, we identified CPB1 correlating with PD severity, which may support clinical staging of PD. In addition, our results showing discrepancy between PEA and ELISA technology suggest that caution should be used when translating proteomic findings into clinical practice. Full article

This paper presents software advances to easily exploit computer architectures consisting of a multi-core CPU and CPU+GPU to accelerate diverse types of high-performance computing (HPC) applications using a single code implementation. The paper describes and demonstrates the performance of the open-source C++ matrix and array (MATAR) library that uniquely offers: (1) a straightforward syntax for programming productivity, (2) usable data structures for data-oriented programming (DOP) for performance, and (3) a simple interface to the open-source C++ Kokkos library for portability and memory management across CPUs and GPUs. The portability across architectures with a single code implementation is achieved by automatically switching between diverse fine-grained parallelism backends (e.g., CUDA, HIP, OpenMP, pthreads, etc.) at compile time. The MATAR library solves many longstanding challenges associated with easily writing software that can run in parallel on any computer architecture. This work benefits projects seeking to write new C++ codes while also addressing the challenges of quickly making existing Fortran codes performant and portable over modern computer architectures with minimal syntactical changes from Fortran to C++. We demonstrate the feasibility of readily writing new C++ codes and modernizing existing codes with MATAR to be performant, parallel, and portable across diverse computer architectures. Full article

Arable land loss has become a critical issue in China because of rapid urbanization, industrial expansion, and unsustainable agricultural practices. While previous studies have explored the factors contributing to this loss, they often fall short in addressing the challenges of spatial heterogeneity and large-scale dataset analysis. This research introduces an innovative approach to geographically weighted regression (GWR) for assessing arable land loss in China, effectively addressing these challenges. Focusing on Chongqing, Guizhou, and Yunnan Provinces over the past two decades, it examines spatial autocorrelation with R-squared values exceeding 0.6 and residuals. Eight factors, including environmental elements (rain, evaporation, slope, digital elevation model) and human activities (distance to city, distance to roads, population, GDP), were analyzed. By visualizing and analyzing R² spatial patterns, the results reveal a clear spatial agglomeration distribution, primarily in urban areas with industries, highly urbanized cities, and flat terrains near rivers, influenced by GDP, population, rain, and slope. The novelty of this study is that it significantly enhances GWR computational capabilities for handling extensive datasets by utilizing Compute Unified Device Architecture (CUDA) on a high-performance GPU cloud server. Simultaneously, it conducts comprehensive analyses of the GWR model’s local results through visualization and spatial autocorrelation tools, enhancing the interpretability of the GWR model. Through spatial clustering analysis of local results, this study enables targeted exploration of factors influencing arable land changes in various temporal and spatial dimensions while also evaluating the reliability of the model results. Full article

Large-scale urban water distribution network simulation plays a critical role in the construction, monitoring, and maintenance of urban water distribution systems. However, during the simulation process, matrix inversion calculations generate a large amount of computational data and consume significant amounts of time, posing challenges for practical applications. To address this issue, this paper proposes a parallel gradient calculation algorithm based on GPU hardware and the CUDA Toolkit library and compares it with the EPANET model and a model based on CPU hardware and the Armadillo library. The results show that the GPU-based model not only achieves a precision level very close to the EPANET model, reaching 99% accuracy, but also significantly outperforms the CPU-based model. Furthermore, during the simulation, the GPU architecture is able to efficiently handle large-scale data and achieve faster convergence, significantly reducing the overall simulation time. Particularly in handling larger-scale water distribution networks, the GPU architecture can improve computational efficiency by up to 13 times. Further analysis reveals that different GPU models exhibit significant differences in computational efficiency, with memory capacity being a key factor affecting performance. GPU devices with larger memory capacity demonstrate higher computational efficiency when processing large-scale water distribution networks. This study demonstrates the advantages of GPU acceleration technology in the simulation of large-scale urban water distribution networks and provides important theoretical and technical support for practical applications in this field. By carefully selecting and configuring GPU devices, the computational efficiency of large-scale water distribution networks can be significantly improved, providing more efficient solutions for future urban water resource management and planning. Full article

A cross-correlation algorithm to obtain the sub-aperture shifts that occur is a crucial aspect of scene-based SHWS (Shack–Hartmann wavefront sensing). However, when the sub-image is partially absent within the atmosphere, the traditional cross-correlation algorithm can easily obtain the wrong shift results. To overcome this drawback, we propose an algorithm based on SURFs (speeded-up-robust features) matching. In addition, to meet the speed required by wavefront sensing, CUDA parallel optimization of SURF matching is carried out using a GPU thread execution model and a programming model. The results show that the shift error can be reduced by more than two times, and the parallel algorithm can achieve nearly ten times the acceleration ratio. Full article

The mode of coordination of copper(II) ions with dopamine (DA, L) in the binary, as well as ternary systems with Ado, AMP, ADP, and ATP (L′) as second ligands, was studied with the use of experimental—potentiometric and spectroscopic (VIS, EPR, NMR, IR)—methods and computational—molecular modeling and DFT—studies. In the Cu(II)/DA system, depending on the pH value, the active centers of the ligand involved in the coordination with copper(II) ions changed from nitrogen and oxygen atoms (CuH(DA)³⁺, Cu(DA)²⁺), via nitrogen atoms (CuH₂(DA)₂⁴⁺), to oxygen atoms at strongly alkaline pH (Cu(DA)₂²⁺). The introduction of L′ into this system changed the mode of interaction of dopamine from oxygen atoms to the nitrogen atom in the hydroxocomplexes formed at high pH values. In the ternary systems, the ML′-L (non-covalent interaction) and ML′HxL, ML′L, and ML′L(OH)x species were found. In the Cu(II)/DA/AMP or ATP systems, mixed forms were formed up to a pH of around 9.0; above this pH, only Cu(II)/DA complexes occurred. In contrast to systems with AMP and ATP, ternary species with Ado and ADP occurred in the whole pH range at a high concentration, and moreover, binary complexes of Cu(II) ions with dopamine did not form in the detectable concentration. Full article

As a signature phenomenon of massive hypervelocity impacts (HVIs) in space, debris clouds provide critical optical information for satellite remote sensing and the assessment of large-scale impacts. However, studies of the optical scattering properties of debris clouds remain limited, and existing vector radiative transfer (VRT) methods struggle to accurately simulate the optical characteristics of these complex scatterers. To address this gap, this paper presents an improved Monte Carlo VRT program (PGS–MC) for multicomponent polydisperse scatterers to precisely evaluate the radiation and polarization characteristics of complex scatterers. Based on the Monte Carlo ray tracing (MCRT) method, our program introduces a particle grouping strategy (PGS) to further emphasize the importance of accounting for optical property discrepancies between different materials and particle sizes, thus significantly improving the fidelity of VRT simulations. Moreover, our program, developed using the compute unified device architecture (CUDA), can be run parallelly on graphics processing units (GPUs), which effectively reduces the computational time. The validation results indicated that the developed PGS–MC program can accurately and efficiently simulate the polarization of complex 3D scatterers. A further investigation showed that the polarization characteristics of debris clouds are highly sensitive to parameters such as the angle between the incident and detection directions, number density, particle size distribution, debris material, and wavelength. In addition, the polarization imaging of debris clouds offers distinct advantages over intensity imaging. This study offers guidance for analyzing the VRT properties of massive HVI debris clouds. Additionally, it provides a practical tool and concrete ideas for modeling the polarization characteristics of various complex scatterers, such as aircraft contrails and clouds, etc. Full article

The digital representation of seafloor, a challenge in UNESCO’s Ocean Decade initiative, is essential for sustainable development support and marine environment protection, aligning with the United Nations’ 2030 program goals. Accuracy in seafloor representation can be achieved through remote sensing measurements, including acoustic and laser sources. Ground truth information integration facilitates comprehensive seafloor assessment. The current seafloor mapping paradigm benefits from the object-based image analysis (OBIA) approach, managing high-resolution remote sensing measurements effectively. A critical OBIA step is the segmentation process, with various algorithms available. Recent artificial intelligence advancements have led to AI-powered segmentation algorithms development, like the Segment Anything Model (SAM) by META AI. This paper presents the SAM approach’s first evaluation for seafloor mapping. The benchmark remote sensing dataset refers to Puck Lagoon, Poland and includes measurements from various sources, primarily multibeam echosounders, bathymetric lidar, airborne photogrammetry, and satellite imagery. The SAM algorithm’s performance was evaluated on an affordable workstation equipped with an NVIDIA GPU, enabling CUDA architecture utilization. The growing popularity and demand for AI-based services predict their widespread application in future underwater remote sensing studies, regardless of the measurement technology used (acoustic, laser, or imagery). Applying SAM in Puck Lagoon seafloor mapping may benefit other seafloor mapping studies intending to employ AI technology. Full article

Differences in field of view may occur during unmanned aerial remote sensing imaging applications with acousto-optic tunable filter (AOTF) spectral imagers using zoom lenses. These differences may stem from image size deformation caused by the zoom lens, image drift caused by AOTF wavelength switching, and drone platform jitter. However, they can be addressed using hyperspectral image registration. This article proposes a new coarse-to-fine remote sensing image registration framework based on feature and optical flow theory, comparing its performance with that of existing registration algorithms using the same dataset. The proposed method increases the structure similarity index by 5.2 times, reduces the root mean square error by 3.1 times, and increases the mutual information by 1.9 times. To meet the real-time processing requirements of the AOTF spectrometer in remote sensing, a development environment using VS2023+CUDA+OPENCV was established to improve the demons registration algorithm. The registration algorithm for the central processing unit+graphics processing unit (CPU+GPU) achieved an acceleration ratio of ~30 times compared to that of a CPU alone. Finally, the real-time registration effect of spectral data during flight was verified. The proposed method demonstrates that AOTF hyperspectral imagers can be used in real-time remote sensing applications on unmanned aerial vehicles. Full article