GPGPU Research Articles

Reducing the replication time for structural estimations: A successful replication of “An Anatomy of International Trade” using GPU computing

AbstractEaton, Kortum, and Kramarz (2011) (EKK) discovered empirical patterns from French manufacturing firms that a baseline firm heterogeneity model could not explain. The authors proposed and estimated a model that closely matches the patterns observed in French data. This paper successfully replicates their findings using author‐provided data, re‐implementing their algorithms in Python and leveraging graphics processing unit computing to significantly boost computational speed. Applying the model to Chinese manufacturing data, despite differences in context, showed that the model explains most observed patterns well.

Real-Time Identification of Strawberry Pests and Diseases Using an Improved YOLOv8 Algorithm

Strawberry crops are susceptible to a wide range of pests and diseases, some of which are insidious and diverse due to the shortness of strawberry plants, and they pose significant challenges to accurate detection. Although deep learning-based techniques to detect crop pests and diseases are effective in addressing these challenges, determining how to find the optimal balance between accuracy, speed, and computation remains a key issue for real-time detection. In this paper, we propose a series of improved algorithms based on the YOLOv8 model for strawberry disease detection. These include improvements to the Convolutional Block Attention Module (CBAM), Super-Lightweight Dynamic Upsampling Operator (DySample), and Omni-Dimensional Dynamic Convolution (ODConv). In experiments, the accuracy of these methods reached 97.519%, 98.028%, and 95.363%, respectively, and the F1 evaluation values reached 96.852%, 97.086%, and 95.181%, demonstrating significant improvement compared to the original YOLOv8 model. Among the three improvements, the improved model based on CBAM has the best performance in training stability and convergence, and the change in each index is relatively smooth. The model is accelerated by TensorRT, which achieves fast inference through highly optimized GPU computation, improving the real-time identification of strawberry diseases. The model has been deployed in the cloud, and the developed client can be accessed by calling the API. The feasibility and effectiveness of the system have been verified, providing an important reference for the intelligent research and application of strawberry disease identification.

Leapfrogging Sycamore: Harnessing 1432 GPUs for 7× faster quantum random circuit sampling

Abstract Random quantum circuit sampling serves as a benchmark to demonstrate quantum computational advantage. Recent progress in classical algorithms, especially those based on tensor network methods, has significantly reduced the classical simulation time and challenged the claim of the first-generation quantum advantage experiments. However, in terms of generating uncorrelated samples, time-to-solution, and energy consumption, previous classical simulation experiments still underperform the Sycamore processor. Here we report an energy-efficient classical simulation algorithm, using 1432 GPUs to simulate quantum random circuit sampling which generates uncorrelated samples with higher linear cross entropy score and is 7 times faster than the Sycamore 53 qubits experiment. We propose a post-processing algorithm to reduce the overall complexity, and integrate state-of-the-art high-performance general-purpose GPU to achieve two orders of lower energy consumption compared to previous works. Our work provides the first unambiguous experimental evidence to refute Sycamore’s claim of quantum advantage, and redefines the boundary of quantum computational advantage using random circuit sampling.

Adaptive random-grid and progressive color secret sharing with CNN super-resolution on a many-core system

This study introduces an innovative method combining Convolutional Neural Networks (CNN) with random grid generation for enhanced visual secret sharing. Our approach supports color images, reduces processing time, and offers non-pixel expansion and flexible share combinations. Utilizing GPU computation, it significantly improves practicality and efficiency by operating in a feed-forward manner, avoiding complex optimization. Experimental results demonstrate superior metrics: maximum PSNR of 32 dB, 99% NCC correlation with benchmarks, 2.9% NAE, and SSIM of 98%. We achieve substantial speedups– 493 × for 256 × 256 and 3145 × for 1024 × 1024 images–compared to sequential models. The scheme exhibits robustness against attacks, evidenced by CMY component and share histogram similarities, and scalability shown in combinatorial explosion visualization. These findings underscore the efficacy and efficiency of our approach, advancing secure image sharing applications significantly.

General heterogeneous real-time baseband backend system scheme applied to QTT based on RFSoC+GPU

This paper proposes a heterogeneous real-time baseband astronomical backend solution centered on RFSoC and combined GPU. The solution implemented the integration of the high-performance, completely configurable RF signal link to digitize, process, and transmit increasingly massive radio astronomy signals and provided a highly flexible and scalable scheme for the QTT backend. The RFSoC-based integration RF pre-processing front-end directly digitizes dual-polarization signals at the sampling rate of 2048 MHz and converts them to dual-16-channel data streams for parallel mixing and polyphase decimation filtering. The generated baseband data with 256 MHz center frequency and 100 MHz bandwidth is packaged and output to the GPU server through the 100GbE Interface. Complete processes of direct RF sampling, pre-processing, and high-speed transmission are implemented at the receiver. The GPU server executes real-time post-processing at the remote end. The dynamic configuration of two observation modes can be achieved with frequency resolutions of 3.051 and 0.763 kHz. The H2CO absorption line observations with the Nanshan 26-m radio telescope verified the system’s performance and the design’s rationality. This paper explores and confirms the RFSoC’s applicability on the backend. The flexible and efficient QTT backend scheme was planned based on the proposed real-time baseband pattern by expanding multi-RFSoC and GPU computing nodes.

Some computational results on a conjecture of de Polignac about numbers of the form p + 2k

Let U be the set of positive odd numbers that can not be written in the form p+2k. Recently, by analyzing possible prime divisors of b, Chen proved b≥11184810 and ω(b)≥7 if an arithmetic progression a(modb) is in U, with ω(b)=7 if and only if b=11184810, where ω(n) is the number of distinct prime divisors of n. In this paper, we take a computational approach to prove b≥11184810 and provide all possible values of a if a(mod11184810) is in U. Moreover, we explicitly construct nontrivial arithmetic progressions a(modb) in U with ω(b)=8, 9, 10, or 11, and provide potential nontrivial arithmetic progressions a(modb) in U such that ω(b)=s for any fixed s≥12. Furthermore, we improve the upper bound estimate of numbers of the form p+2k by Habsieger and Roblot in 2006 to 0.490341088858244 by enhancing their algorithm and employing GPU computation.

Design and implementation of a scalable correlator based on ROACH2 + GPU cluster for tianlai 96-dual-polarization antenna array

The digital correlator is one of the most crucial data processing components of a radio telescope array. With the scale of radio interferometeric array growing, many efforts have been devoted to developing a cost-effective and scalable correlator in the field of radio astronomy. In this paper, a 192-input digital correlator with six CASPER ROACH2 boards and seven GPU servers has been deployed as the digital signal processing system for Tianlai cylinder pathfinder located in Hongliuxia observatory. The correlator consists of 192 input signals (96 dual-polarization), 125-MHz bandwidth, and full-Stokes output. The correlator inherits the advantages of the CASPER system, for example, low cost, high performance, modular scalability, and a heterogeneous computing architecture. With a rapidly deployable ROACH2 digital sampling system, a commercially expandable 10 Gigabit switching network system, and a flexible upgradable GPU computing system, the correlator forms a low-cost and easily-upgradable system, poised to support scalable large-scale interferometeric array in the future.

Wave Propagation for Structural Health Monitoring (WaveProSHM): open software with GUI

Guided wave–based SHM methods have been explored extensively in recent decades. The wave propagation modelling including piezoelectric transducers and various damage types helps understand intricacies of guided wave behavior and for designing a robust SHM system. However, modelling aspects are still challenging. Moreover, commercially available software for the modelling of guided waves often lacks appropriate high-order elements and code optimization for fast GPU computations. This research aimed to develop and disseminate to the SHM community the open software based on highly efficient vectorised code of the time domain spectral element method which can be run on GPU. The developed code is written in MATLAB with Graphical User Interface (GUI) and communicates with GMSH – open source finite element mesh generator. It can model signals of propagating guided waves in structures which are excited and registered by an array of piezoelectric transducers. The capabilities of the software are shown in an example of a large-scale problem of interacting guided waves with delamination.

Deep learning for 3D cephalometric landmarking with heterogeneous multi-center CBCT dataset.

Cephalometric analysis is critically important and common procedure prior to orthodontic treatment and orthognathic surgery. Recently, deep learning approaches have been proposed for automatic 3D cephalometric analysis based on landmarking from CBCT scans. However, these approaches have relied on uniform datasets from a single center or imaging device but without considering patient ethnicity. In addition, previous works have considered a limited number of clinically relevant cephalometric landmarks and the approaches were computationally infeasible, both impairing integration into clinical workflow. Here our aim is to analyze the clinical applicability of a light-weight deep learning neural network for fast localization of 46 clinically significant cephalometric landmarks with multi-center, multi-ethnic, and multi-device data consisting of 309 CBCT scans from Finnish and Thai patients. The localization performance of our approach resulted in the mean distance of 1.99 ± 1.55 mm for the Finnish cohort and 1.96 ± 1.25 mm for the Thai cohort. This performance turned out to be clinically significant i.e., ≤ 2 mm with 61.7% and 64.3% of the landmarks with Finnish and Thai cohorts, respectively. Furthermore, the estimated landmarks were used to measure cephalometric characteristics successfully i.e., with ≤ 2 mm or ≤ 2° error, on 85.9% of the Finnish and 74.4% of the Thai cases. Between the two patient cohorts, 33 of the landmarks and all cephalometric characteristics had no statistically significant difference (p < 0.05) measured by the Mann-Whitney U test with Benjamini-Hochberg correction. Moreover, our method is found to be computationally light, i.e., providing the predictions with the mean duration of 0.77 s and 2.27 s with single machine GPU and CPU computing, respectively. Our findings advocate for the inclusion of this method into clinical settings based on its technical feasibility and robustness across varied clinical datasets.

Advances in granular flow modeling: GPU-based multi-sphere DEM approach and tumbling mill dynamics

Modeling irregular-shaped particulate flow systems poses challenges, especially in mineral processing where accurate representation of particle shape is crucial for capturing macroscopic behavior. The multi-sphere approach, employing DEM, addresses this by approximating particles as overlapping spheres within GPU computing. This study validates particle flow dynamics using literature-based data, showcasing the model’s fidelity and GPU advantages. However, calibration and shape approximation limitations require careful consideration. A quaternion-based orientation method enhances DEM and validates experimental data from packing, hoppers, and tumbling mills. Simulations reveal how particle shape influences granular dynamics, affecting lift-off tendencies, responses to lifter face angles, and power draw trends with mill speed changes. Cubical and spherical particles exhibit orderly packing behavior, contrasting with irregular shapes. This investigation underscores the significant role of particle shape in granular dynamics and its implications for industrial processes.

P‐239: Late‐News Poster: Modeling Particle Deposition in Evaporating Colloidal Droplets

Inkjet printing is a cost‐effective method of particle deposition applicable, for example, in the manufacture of Quantum Dot (QD) color filters. However, it is a challenge to achieve uniform deposition of particles because evaporating colloidal dispersions are prone to the coffee‐ring effect.When the particle size is small, it is difficult to gauge directly by experiment how particles migrate during the evaporation process. Consequently, it is a challenge to optimize the process conditions to improve uniformity by experiment alone. To aid in this task, we have developed a simulator for colloidal droplet evaporation. Rather than use an effective particle concentration field as in previous works, motion of each individual particle is calculated by Stokesian Dynamics, with particle interactions governed by the DLVO theory. In order to facilitate the simulation of large collections of particles, GPU computing is applied to reduce the simulation time.A phase‐field approach is taken to capture the interface physics at the solvent‐air interface. Within a colloidal droplet we find that, by controlling the temperature, it is possible to promote Marangoni flow and, in turn, improve deposition uniformity. In future, we aim to use simulation to further optimize the process conditions, such as the chamber pressure and temperature, in order maximize uniformity.

The Galaxy platform for accessible, reproducible, and collaborative data analyses: 2024 update.

Galaxy (https://galaxyproject.org) is deployed globally, predominantly through free-to-use services, supporting user-driven research that broadens in scope each year. Users are attracted to public Galaxy services by platform stability, tool and reference dataset diversity, training, supportand integration, which enables complex, reproducible, shareable data analysis. Applying the principles of user experience design (UXD), has driven improvements in accessibility, tool discoverability through Galaxy Labs/subdomains, and a redesigned Galaxy ToolShed. Galaxy tool capabilities are progressing in two strategic directions: integrating general purpose graphical processing units (GPGPU) access for cutting-edge methods, and licensed tool support. Engagement with global research consortia is being increased by developing more workflows in Galaxy and by resourcing the public Galaxy services to run them. The Galaxy Training Network (GTN) portfolio has grown in both size, and accessibility, through learning paths and direct integration with Galaxy tools that feature in training courses. Code development continues in line with the Galaxy Project roadmap, with improvements to job scheduling and the user interface. Environmental impact assessment is also helping engage users and developers, reminding them of their role in sustainability, by displaying estimated CO2 emissions generated by each Galaxy job.

Knowledge Graph Reasoning via Dynamic Subgraph Attention with Low Resource Computation

Knowledge graphs (KGs) suffer from inherent incompleteness, which has spurred research into knowledge graph reasoning (KGR), i.e., ways to infer missing facts based on existing triples. The most prevalent approaches employ a static mechanism for entity representation across the entire graph, sharing entity embeddings for different query relations. Nevertheless, these static entity embeddings fail to capture specific semantics for various query scenarios, resulting in inaccurate entity representations and susceptibility to reasoning errors. Furthermore, the whole graph learning style requires substantial computational resources, especially since KGs are often of large-scale. Consequently, these methods are impractical in low-resource environments. To address these issues, we devise a framework called dynamic subgraph attention (DSA), which learns dynamic entity embeddings for different query relations across subgraphs. In our approach, by utilizing multi-hop path history information obtained through path-based learning, we guide a dynamic attention mechanism to generate dynamic entity embeddings for different query relations. To ensure the semantic information of entities, the embedding-based and path-based learning are jointly trained for KGR. Additionally, the proposed dynamic aggregation mechanism operates on subgraphs, resulting in a remarkable 6–7 times resource conservation compared to GPU computations. Empirical experiments further demonstrate that DSA outperforms the current methods significantly on three benchmark datasets.

AI Face Recognition and Processing Technology Based on GPU Computing

In recent years, with the development of deep neural network technology, real-time object detection has become increasingly common in mobile applications. However, practical application requirements drive the algorithm to optimize in terms of speed, energy consumption and accuracy. This paper introduces the application of artificial intelligence in the field of face recognition, especially using TensorRT accelerated reasoning technology to improve the speed and performance of face recognition. At the same time, the paper also discusses the key role of GPU computing in face recognition, and expounds the importance of AI chips for optimizing inference tasks. Through the analysis of experimental results and methods, the performance advantages and application prospects of BlazeFace algorithm in mobile applications are demonstrated, which provides a valuable reference for industry.

Exploring memory synchronization and performance considerations for FPGA platform using the high-abstracted OpenCL framework: Benchmarks development and analysis.

A key benefit of the Open Computing Language (OpenCL) software framework is its capability to operate across diverse architectures. Field programmable gate arrays (FPGAs) are a high-speed computing architecture used for computation acceleration. This study investigates the impact of memory access time on overall performance in general FPGA computing environments through the creation of eight benchmarks within the OpenCL framework. The developed benchmarks capture a range of memory access behaviors, and they play a crucial role in assessing the performance of spinning and sleeping on FPGA-based architectures. The results obtained guide the formulation of new implementations and contribute to defining an abstraction of FPGAs. This abstraction is then utilized to create tailored implementations of primitives that are well-suited for this platform. While other research endeavors concentrate on creating benchmarks with the Compute Unified Device Architecture (CUDA) to scrutinize the memory systems across diverse GPU architectures and propose recommendations for future generations of GPU computation platforms, this study delves into the memory system analysis for the broader FPGA computing platform. It achieves this by employing the highly abstracted OpenCL framework, exploring various data workload characteristics, and experimentally delineating the appropriate implementation of primitives that can seamlessly integrate into a design tailored for the FPGA computing platform. Additionally, the results underscore the efficacy of employing a task-parallel model to mitigate the need for high-cost synchronization mechanisms in designs constructed on general FPGA computing platforms.

ShaderPerFormer: Platform-independent Context-aware Shader Performance Predictor

The ability to model and predict the execution time of GPU computations is crucial for real-time graphics application development and optimization. While there are many existing methodologies for graphics programmers to provide such estimates, those methods are often vendor-dependent, require the platforms to be tested, or fail to capture the contextual influences among shader instructions. To address this challenge, we propose ShaderPerFormer, a platform-independent, context-aware deep-learning approach to model GPU performance and provide end-to-end performance predictions on a per-shader basis. To provide more accurate predictions, our method contains a separate stage to gather platform-independent shader program trace information. We also provide a dataset consisting of a total of 54,667 fragment shader performance samples on 5 different platforms. Compared to the PILR and SH baseline methods, our approach reduces the average MAPE across five platforms by 8.26% and 25.25%, respectively.

Rapid calculation for avalanche maps by GPGPU-based snow avalanche model

The required time for producing snow avalanche maps is influenced by computation speed of simulations. Commonly, integrating terrain assessment with dynamic flow simulation aids in mapping dangerous areas for human and structural threats. This approach enables the evaluation of avalanche paths, as well as the assessment of flow rate and thickness during avalanche movement. However, the substantial computational cost of the simulation results in long calculation times when using the Central Processing Unit (CPU). In this study, a new rapid snow avalanche simulator was developed by applying massively parallel computation with the General-Purpose computing on Graphics Processing Unit (GPGPU) technique. By avoiding slower data transfer and utilizing faster memory, computational speed could be accelerated up to 80 times faster than conventional simulation using a CPU. Additionally, the rapid calculation models were validated based on the Mt. Nasu event in 2017, and pilot studies of the avalanche map of Mt. Nasu in Japan demonstrated the usefulness of the developed model for vulnerability evaluation. A total of 123 simulations were conducted for each susceptible source area, and all simulations were completed within only 6.5 h. This high-performance calculation can significantly reduce the time cost of producing and expanding conventional avalanche maps.

Expression of Concern: 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

Expression of Concern: 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

Neural networks as effective surrogate models of radio-frequency quadrupole particle accelerator simulations

Radio-frequency quadrupoles (RFQs) are multi-purpose linear particle accelerators that simultaneously bunch and accelerate charged particle beams. They are ubiquitous in accelerator physics, especially as injectors to higher-energy machines, owing to their impressive efficiency. The design and optimization of these devices can be lengthy due to the need to repeatedly perform high-fidelity simulations. Several recent papers have demonstrated that machine learning can be used to build surrogate models (fast-executing replacements of computationally costly beam simulations) for order-of-magnitude computing time speedups. However, while these pilot studies are encouraging, there is room to improve their predictive accuracy. Particularly, beam summary statistics such as emittances (an important figure of merit in particle accelerator physics) have historically been challenging to predict. For the first time, we present a surrogate model trained on 200 000 samples that yields < 6% mean average percent error for the predictions of all relevant beam output parameters from defining RFQ design parameters, solving the problem of poor emittance predictions by identifying and including hidden variables which were not accounted for previously. These surrogate models were made possible by using the Julia language and GPU computing; we briefly discuss both. We demonstrate the utility of surrogate modeling by performing a multi-objective optimization using our best model as a callback in the objective function to select an optimal RFQ design. We consider trade-offs in RFQ performance for various choices of Pareto-optimal design variables—common issues for any multi-objective optimization scheme. Lastly, we make recommendations for input data preparation, selection, and neural network architectures that pave the way for future development of production-capable surrogate models for RFQs and other particle accelerators.

Real-time soft body dissection simulation with parallelized graph-based shape matching on GPU

Background and ObjectiveInteractive soft tissue dissection has been a fundamental procedure in virtual surgery systems. Existing cutting algorithms involve complex topology changes of simulation meshes, which can increase simulation overhead and produce visual artifacts. In this paper, we proposed a novel graph-based shape-matching method that allows for real-time, flexible, progressive, and discontinuous cuts on soft tissue. MethodsWe employed shape-matching constraints within the position-based dynamics (PBD) framework, a widely adopted approach for real-time simulation applications. The soft tissue was effectively modeled using overlapping clusters, each governed by shape-matching constraints. The dissection process was bifurcated into two distinct stages. In the first stage, the surgical scalpel presses the surface of the soft tissue. The soft tissue is cut apart when the surface pressure exceeds a threshold, entering the second stage. To address the discrepancy between the visual mesh and the simulation model during cluster separation, we developed an Aggregate Finding Connected Components (AFCC) algorithm, optimized for GPU computation and integrated with a background grid. This approach also avoids ghost forces and fragmentation artifacts. To control the increase in the number of clusters, we also propose a merging strategy that can run in parallel. ResultsOur simulation outcomes demonstrated that the AFCC dissection algorithm effectively manages cluster separation and expansion with robustness. There were no ghost forces between the cutting surface and unrealistic fragments. Our simulation capability extended to supporting intricate and discontinuous cutting routes. Our dissection simulation maintained real-time performance even with over 100,000 particles constituting the soft tissue. ConclusionsOur real-time and robust surgical dissection simulation technique enables the performance of complex cuts in various surgical scenarios, demonstrating its potential in virtual surgery applications.