Supercomputer Research Articles

A Numerical Study on the Flow Field and Classification Performance of an Industrial-Scale Micron Air Classifier under Various Outlet Mass Airflow Rates

In this study, the gas−particle flow field in a real-size industrial-scale micron air classifier manufactured by Phenikaa Group using 3D transient simulations with the FWC-RSM–DPM (Four-Way Coupling-Reynold Stress Model-Discrete Phase Model) in ANSYS Fluent 2022 R2 and with the assistance of High-Performance Computing (HPC) systems is explored. A comparison among three coupling models is carried out, highlighting the significant influence of the interactions between solid and gas phases on the flow field. The complex two-phase flow, characterized by the formation of multiple vortices with different sizes, positions, and rotation directions, is successfully captured on the real-size model of the classifier. Additionally, analyzing the effects of the vortices on the flow field provides a comprehensive understanding of the gas–solid flow field and the classification mechanism. The effect of the outlet mass airflow rate is also investigated. The classifier’s Key Performance Indicators (KPIs: d50, K, η, ΔP) and the constrained condition of the particle size distribution curve of the final product are used to evaluate the classification efficiency. The contributions of this work are as follows: (i) a simulation analysis of a real-size industrial-scale classifier is conducted that highlights its advantages over a lab-scale one; (ii) a comparison is conducted among three coupling models, showing the advancement of four-way coupling in providing accurate results for simulations of interactions between the gas phase and particles; and (iii) the particle size distribution curve performances of a classified product under different simulation models and outlet airflow rates are addressed, from which optimal parameters can be selected in the design and operation processes to achieve the required efficiency of an air classifier.

Enabling high-performance cloud computing for the Community Multiscale Air Quality Model (CMAQ) version 5.3.3: performance evaluation and benefits for the user community.

The Community Multiscale Air Quality Model (CMAQ) is a local- to hemispheric-scale numerical air quality modeling system developed by the U.S. Environmental Protection Agency (USEPA) and supported by the Community Modeling and Analysis System (CMAS) center. CMAQ is used for regulatory purposes by the USEPA program offices and state and local air agencies and is also widely used by the broader global research community to simulate and understand complex air quality processes and for computational environmental fate and transport and climate and health impact studies. Leveraging state-of-the-science cloud computing resources for high-performance computing (HPC) applications, CMAQ is now available as a fully tested, publicly available technology stack (HPC cluster and software stack) for two major cloud service providers (CSPs). Specifically, CMAQ configurations and supporting materials have been developed for use on their HPC clusters, including extensive online documentation, tutorials and guidelines to scale and optimize air quality simulations using their services. These resources allow modelers to rapidly bring together CMAQ, cloud-hosted datasets, and visualization and evaluation tools on ephemeral clusters that can be deployed quickly and reliably worldwide. Described here are considerations in CMAQ version 5.3.3 cloud use and the supported resources for each CSP, presented through a benchmark application suite that was developed as an example of a typical simulation for testing and verifying components of the modeling system. The outcomes of this effort are to provide findings from performing CMAQ simulations on the cloud using popular vendor-provided resources, to enable the user community to adapt this for their own needs, and to identify specific areas of potential optimization with respect to storage and compute architectures.

Robust and computationally efficient design for run-of-river hydropower

This paper introduces innovative approaches for robust and computationally efficient optimal design of run-of-river hydropower plants. Compared with existing design software, it (1) integrates optimized turbine operations into design optimization instead of following predefined operational rules, and (2) combines this with a regular sampling of the flow duration curve to significantly reduce data inputs. Our rigorous benchmarking demonstrates that (1) operation optimization improves design performance at low computational cost, whilst (2) data input reduction slashes computational costs by over 92% with minimal impact on design recommendations and key robustness analysis insights. Taken together, these innovations make integrated design and operation optimization, complete with in-depth robustness analysis, laptop-accessible. They also reinforce sustainability efforts by minimizing the need for high-performance computing and large associated embodied greenhouse gas emissions.

Research on Multiplication Routine Based on Reconfigurable Four-Valued Logic Processor

Despite the indispensable role of traditional electronic computers in modern society, their limitations in parallel processing capabilities, bit-width constraints, and processor bit-width are becoming increasingly apparent, especially when handling large-scale datasets and complex computational tasks. Although hardware technology and algorithm optimization continue to advance, the arithmetic units of traditional computers—adders—remain constrained by carry delay and bit-width limitations. This bottleneck is particularly pronounced in multiplication operations, mainly when adders are used for partial product accumulation. However, since 2018, the emergence of a new type of Reconfigurable Four-Valued Logic Electronic Processor (RFLEP) has provided a potential solution to these traditional limitations. With its large processor bit-width, flexible bit grouping capabilities, and dynamic hardware function reconfiguration features, this processor has brought revolutionary changes to the field of computing. In this context, this paper proposes and implements a Reconfigurable Four-Valued Logic Multiplication Routine (RFLMR) tailored explicitly for the RFLEP. The RFLMR utilizes the Modified Signed-Digit (MSD) representation method in multi-valued logic combined with the M transformation in four-valued logic to generate partial products. These partial products are then efficiently summed in parallel using the JW-MSD parallel adder, achieving the rapid execution of multiplication operations. Experimental results demonstrate that the multiplication routine based on the RFLEP performs multiplication operations accurately and meets theoretical expectations regarding implementation efficiency and performance. This research not only provides new ideas for developing next-generation high-performance computing systems but also paves the way for exploring more efficient and powerful computing models, heralding a profound transformation in future computing technology.

Mathematical Modeling of Electrical Power Systems for Fault Detection

Control networks are critical offices that have to be exceptionally solid and productive. Finding issues in these frameworks is exceptionally imperative for keeping them steady, keeping gear from breaking, and maintaining a strategic distance from long power blackouts. It has gotten to be clear that scientific modeling may be a awesome way to think about and discover issues in control frameworks. Taking after this strategy requires making a scientific show of the control framework, which incorporates motors, transformers, transmission lines, and loads. A few physical laws, like Ohm's Law and Kirchhoff's laws, are frequently utilized within the models to accurately show how electrical circuits work when they are working legitimately and when they aren't. On the off chance that you need to discover issues, you'll utilize numerical models to form flaws like brief circuits, open circuits, and ground deficiencies appear genuine. Through examining how the framework responds to these fake issues, engineers can make programs that can discover genuine flaws and tell where they are when they happen. These calculations frequently utilize strategies like evaluating the state, evaluating the parameters, and recognizing designs. Utilizing real-time information to figure the current state of the framework is what state estimation is all approximately. On the other hand, parameter estimation is all almost finding changes in framework parameters that might cruel there is a problem. Design acknowledgment employments past information and machine learning to discover peculiarities and sort diverse sorts of flaws. Differential conditions, framework polynomial math, and numerical strategies are fair some of the progressed science instruments that are frequently utilized to unravel the complicated conditions that depict how the power framework works. Putting these models beside unused computer innovations like high-performance computing and real-time simulations makes issue location strategies more exact and quicker. As green vitality sources and shrewd network advances are included to control frameworks, they will continue to alter. Mathematical modeling will become more and more important to make sure they work safely and efficiently. It talks about how to mathematically model electrical power systems to find faults, what they can be used for, and what the future holds for this field.

A novel approach for population-receptive field mapping using high-performance computing

A novel approach for population-receptive field mapping using high-performance computing

Application of high-performance computing in sustainable energy campus building design

In high-performance computing (HPC) systems, energy and power considerations play an increasingly important role. This work aims to ensure the implementation of China’s green and ecological concepts, respond to China’s strategy of building an environment-friendly and resource-saving society, and actively promote the construction of sustainable development campuses. First, the theoretical basis of sustainable energy campus architecture is described. Next, the teaching feedback model under HPC is constructed. Finally, with the evaluation results of students’ task completion processes and students’ task works as measurement indexes, the corresponding data is collected and comprehensively evaluated and analyzed to verify the effectiveness of the model. The analysis results indicate that most students are able to complete tasks within two hours; however, their proficiency with multimedia technology is inadequate, and they lack initiative in the learning process. There is a correlation between the overall evaluation of task performance and the students’ level of understanding of the tasks. By implementing a teaching feedback model, students’ learning enthusiasm and the quality of their work improved, providing effective educational support for promoting sustainable development on campus. Overall, the awareness of using computers and other multimedia technologies among students is not strong, and their learning process is relatively closed, with insufficient enthusiasm and initiative. This model can achieve the acquisition, integration, and statistical analysis of teacher feedback information. The model can realize the acquisition, integration, and statistical analysis of teachers’ feedback information. This work hopes to use this learning and feedback mode in practical teaching to address specific problems in computer multimedia courses.

Environmental and social performance relationships to firm efficiency: Evidence from the semiconductor industry

Escalating global demand for high-performance computing and smart devices has driven rapid expansion in the semiconductor industry. This expansion raises environmental and societal sustainability concerns among stakeholders. Although there is a growing commitment to environmental and social responsibilities by semiconductor companies, the impact of these responsibilities and initiatives on corporate performance remains unclear—especially with respect to general firm operational efficiency. Thus, this study employs a sample of 129 listed Chinese semiconductor companies and a slack-based data envelopment analysis (SBM-DEA) methodology to examine the relationship between corporate environmental and social performance and firm efficiency. We also investigate how supply chain position—an important industrial characteristic—moderates these relationships. The findings reveal that: (1) China's semiconductor industry efficiency is relatively low, particularly in the chip design sub-sector; (2) there is a positive relationship between corporate environmental performance and firm efficiency; and (3) supply chain position moderates the relationship between both corporate environmental and social performance and firm efficiency. These findings offer important implications for managers and researchers aiming to foster greater semiconductor industry sustainable development.

V2X Communication Computing Resource Collaboration Technology Based on Deep Reinforcement Learning

In traditional cloud computing, data needs to be transferred from distributed data sources to remote cloud data centers for high-performance computing. However, tasks often experience network transmission delays of hundreds of milliseconds or more from data sources to cloud data centers, which does not meet the need for low latency in V2X (Vehicle to Everything) systems. On the basis of the existing heterogeneous task scenario, this paper proposes a communication computing resource collaboration technology scheme based on heterogeneous task, which can meet the requirements of delay and cost when facing tasks with different delay requirements and different divisibility. This scheme uses the communication computing resources of the roadside unit and the service vehicle, so that the computing task of the task vehicle can be completed within the delay requirement and the cost is minimal. We use the deep reinforcement learning method to study, and finally the performance of the proposed scheme is verified by simulation.

Gels for Water Remediation: Current Research and Perspectives.

The development of cost-effective and high-performance technologies for wastewater treatment is essential for achieving a sustainable economy. Among the various methods available for water remediation, adsorption is widely recognized as an effective and straightforward approach for removing a range of pollutants. Gel materials, particularly hydrogels and aerogels, have attracted significant research interest due to their unique properties. Hydrogels, for instance, are noted for their ability to be regenerated and reused, ease of separation and handling, and suitability for large-scale applications. Additionally, their low cost, high water absorption capacity, and contribution to environmental protection are important advantages. Aerogels, on the other hand, are distinguished by their low thermal conductivity, transparency, flexibility, high porosity, mechanical strength, light weight, large surface area, and ultralow dielectric constant. This review provides a comprehensive analysis of the current literature, highlighting gaps in knowledge regarding the classification, preparation, characterization, and key properties of these materials. The potential application of hydrogels and aerogels in water remediation, particularly in removing contaminants such as dyes, heavy metals, and various organic and inorganic pollutants, is also discussed.

In-line rate encrypted links using pre-shared post-quantum keys and DPUs

The development of quantum computers poses a risk to the cryptographic algorithms utilized by current public key infrastructure (PKI). High performance computing (HPC), inter- and intra-datacenter communications, and governmental communications are particularly vulnerable to security attacks by quantum computers. In response, there are on-going efforts on implementing post-quantum cryptography (PQC) despite the considerable overhead involved in processing PQC encrypted data packets in comparison with their classical counterparts. This work aims to help with the transition to quantum resistant cryptographic schemes by means of providing a thorough experimental performance comparison of the four algorithms selected by the National Institute of Standards and Technology (NIST) for standardization when implemented in a data processing unit (DPU). The experiments are conducted at line-rate, demonstrating for the first time operation at almost 100 Gbit/s and hence suitability for high-capacity data center environments.

Thermal Interaction and Cooling of Electronic Device with Chiplet 2.5D Integration

With the development of artificial intelligence (AI) and high-performance computing (HPC), the microelectronic industry is challenged with increased device integration density. Chiplet architecture can break through a variety of limitations on the system-on-chip (SoC) to create a high-computility system. However, chiplet heterogeneous integration suffers from high heat flux and serious thermal interaction problems. The factors affecting thermal interaction are not clear. In this paper, a collective parameter model and a distribution parameter model are developed to clarify the optimization method to mitigate thermal interaction. The trends predicted by the parameter model are consistent with the finite element method (FEM) simulation results. Furthermore, to cool the chiplets, a thermal test vehicle is designed and fabricated, and the cooling performance of the test vehicle with different flow rates, different TIMs (Thermal Interfacial Materials) (DOW5888 vs. liquid metal), and different heat modes is experimentally investigated. Compared with DOW5888, the utilization of liquid metal TIM can mitigate thermal interaction by 56% and 76% at flow rates of 0.2 L/min and 0.8 L/min, respectively. Consequently, at a temperature rise of 60 °C and a flow rate of 0.8 L/min, using liquid metal TIM can achieve heat fluxes of 330 W/cm2 and 520 W/cm2 for two chiplets, respectively.

Viewpoint-dependent lighting on polygonal holograms using bump mapping.

Holograms can be observed from different viewpoints, because light waves can be encoded to propagate in multiple directions. Thus, accurate holograms for 3D display should model viewpoint-dependent light reflections. We proposed a new, to the best of our knowledge, hologram generation method for objects represented by polygonal meshes, whose lighting changes as the viewer moves, all while rendering smooth shading using low-poly objects. The proposed method leverages bump mapping and converts it into a bump-phase map encoding the propagation frequency and then spreads the reflected light wave so that only a specific viewpoint can receive them. Simulation experiments with small pixel pitches confirm the method's high computational performance.

A Novel Method for Optimising Energy Efficiency in High-Performance Computing Systems

This paper introduces a new method to promote energy efficiency in computing systems. The rapidly growing demand for computational power in high-performance computing (HPC) systems is accompanied by a significant increase in energy consumption. This research investigates the many ways to minimise energy consumption in HPC systems without sacrificing computational performance. Building upon previous research we combined and refined ideas to develop an optimised approach. Our method utilises a tri-modular framework incorporating an Energy-Efficient Hardware Design, Resource Management, and Optimisation, and Server Virtualisation. The first module employs a method of designing smart energy-efficient hardware. This method incorporates leakage reduction techniques such as clock gating and sleep states. The second module uses a technique focused on optimising server software. This method is based on Dynamic Voltage and frequency scaling (DVFS) for power management in data centres. The third module is based on server virtualisation and employs software to create multiple virtual machines on a single physical server allowing for a significant reduction in energy and hardware costs. The methods used in these three modules are systematically integrated to produce a more efficient consumption of electricity. This will allow for the computing system to minimise energy consumption without compromising computational power.

On chip control and detection of complex SPP and waveguide modes based on plasmonic interconnect circuits

Abstract Optical interconnects, leveraging surface plasmon modes, are revolutionizing high-performance computing and AI, overcoming the limitations of electrical interconnects in speed, energy efficiency, and miniaturization. These nanoscale photonic circuits integrate on-chip light manipulation and signal conversion, marking significant advancements in optoelectronics and data processing efficiency. Here, we present a novel plasmonic interconnect circuit, by introducing refractive index matching layer, the device supports both pure SPP and different hybrid modes, allowing selective excitation and transmission based on light wavelength and polarization, followed by photocurrent conversion. We optimized the coupling gratings to fine-tune transmission modes around specific near-infrared wavelengths for effective electrical detection. Simulation results align with experimental data, confirming the device’s ability to detect complex optical modes. This advancement broadens the applications of plasmonic interconnects in high-speed, compact optoelectronic and sensor technologies, enabling more versatile nanoscale optical signal processing and transmission.

Application of high performance computing and deep neural network learning in intelligent city mechanical product recommendation

Application of high performance computing and deep neural network learning in intelligent city mechanical product recommendation

Materials Acceleration Platforms (MAPs): Accelerating Materials Research and Development to Meet Urgent Societal Challenges.

Climate Change and Materials Criticality challenges are driving urgent responses from global governments. These global responses drive policy to achieve sustainable, resilient, clean solutions with Advanced Materials (AdMats) for industrial supply chains and economicprosperity. The research landscape comprising industry, academe, and government identified a critical path to accelerate the Green Transition far beyond slow conventional research through Digital Technologies that harness Artificial Intelligence, Smart Automation and High Performance Computing through Materials Acceleration Platforms, MAPs. In this perspective, following the short paper, a broad overview about the challenges addressed, existing projects and building blocks of MAPs will be provided while concluding with a review of the remaining gaps and measures to overcome them.

Non-smooth Bayesian optimization in tuning scientific applications

Tuning algorithmic parameters to optimize the performance of large, complicated computational codes is an important problem involving finding the optima and identifying regimes defined by non-smooth boundaries in black-box functions. Within the Bayesian optimization framework, the Gaussian process surrogate model produces smooth mean functions, but functions in the tuning problem are often non-smooth, which is exacerbated by the fact that we usually have limited sequential samples from the black-box function. Motivated by these issues encountered in tuning, we propose a novel Gaussian process model called a clustered Gaussian process (cGP), where the components are dynamically updated by clustering. In our studies, the performance of cGP can be better than stationary GPs in nearly 90% of the experiments and better than non-stationary GPs in nearly 70% of the repeated experiments while requiring less computational cost. cGP provides a novel approach for dynamic GP, computes more efficiently than recursive partitioning, and discovers non-smoothness regimes. We provide extensive experiments including high-performance computing (HPC) and industrial simulation functions to show the effectiveness of our methods.

Large language model evaluation for high‐performance computing software development

AbstractWe apply AI‐assisted large language model (LLM) capabilities of GPT‐3 targeting high‐performance computing (HPC) kernels for (i) code generation, and (ii) auto‐parallelization of serial code in C ++, Fortran, Python and Julia. Our scope includes the following fundamental numerical kernels: AXPY, GEMV, GEMM, SpMV, Jacobi Stencil, and CG, and language/programming models: (1) C++ (e.g., OpenMP [including offload], OpenACC, Kokkos, SyCL, CUDA, and HIP), (2) Fortran (e.g., OpenMP [including offload] and OpenACC), (3) Python (e.g., numpy, Numba, cuPy, and pyCUDA), and (4) Julia (e.g., Threads, CUDA.jl, AMDGPU.jl, and KernelAbstractions.jl). Kernel implementations are generated using GitHub Copilot capabilities powered by the GPT‐based OpenAI Codex available in Visual Studio Code given simple <kernel> + <programming model> + <optional hints> prompt variants. To quantify and compare the generated results, we propose a proficiency metric around the initial 10 suggestions given for each prompt. For auto‐parallelization, we use ChatGPT interactively giving simple prompts as in a dialogue with another human including simple “prompt engineering” follow ups. Results suggest that correct outputs for C++ correlate with the adoption and maturity of programming models. For example, OpenMP and CUDA score really high, whereas HIP is still lacking. We found that prompts from either a targeted language such as Fortran or the more general‐purpose Python can benefit from adding language keywords, while Julia prompts perform acceptably well for its Threads and CUDA.jl programming models. We expect to provide an initial quantifiable point of reference for code generation in each programming model using a state‐of‐the‐art LLM. Overall, understanding the convergence of LLMs, AI, and HPC is crucial due to its rapidly evolving nature and how it is redefining human‐computer interactions.

Evaluating ARM and RISC-V Architectures for High-Performance Computing with Docker and Kubernetes

This paper thoroughly assesses the ARM and RISC-V architectures in the context of high-performance computing (HPC). It includes an analysis of Docker and Kubernetes integration. Our study aims to evaluate and compare these systems’ performance, scalability, and practicality in a general context and then assess the impact they might have on special use cases, like HPC. ARM-based systems exhibited better performance and seamless integration with Docker and Kubernetes, underscoring their advanced development and effectiveness in managing high-performance computing workloads. On the other hand, despite their open-source architecture, RISC-V platforms presented considerable intricacy and difficulties in working with Kubernetes, which hurt their overall effectiveness and ease of management. The results of our study offer valuable insights into the practical consequences of implementing these architectures for HPC, highlighting ARM’s preparedness and the potential of RISC-V while acknowledging the increased complexity and significant trade-offs involved at this point.