Multi-core Research Articles

Energy-efficient joint performance optimization of cloud data centre users/operator using memetic algorithm

ABSTRACT The use of cloud computing as a cost-effective and flexible method has currently drawn the attention of cloud service providers. When offering various cloud services to users, a certain level of Service Level Agreement (SLA) must be guaranteed by the cloud operator, based on the service received at the user level. Considering the non-linear dependency of network users’ Quality of Experience (QoE) on cloud resources (RAM, disk memory, network bandwidth, and CPU core usage) as well as the non-linear and time-varying dependency of cloud operator efficiency on the level of resources consumed by users, joint optimization of user experience and cloud operator efficiency poses significant computational challenges. This paper focuses on the green optimization of both user experience and the benefits of cloud data centre operators, employing a metaheuristic algorithm, which essentially combines a genetic algorithm with an innovative approach. In this approach, the efficiency of the cloud data centre operator, energy consumption of data centres, and user experience quality are jointly optimized. Simulation results demonstrate the improvement of the combined user/operator utility function compared to traditional methods, considering energy consumption constraints.

Exact parallelized dynamic mode decomposition with Hankel matrix for large-scale flow data

An exact parallel algorithm of dynamic mode decomposition (DMD) with Hankel matrices for large-scale flow data is proposed. The proposed algorithm enables the DMD and the Hankel DMD for large-scale data obtained by high-fidelity flow simulations, such as large-eddy simulations or direct numerical simulations using more than a billion grid points, on parallel computations without any approximations. The proposed algorithm completes the computations of the DMD by utilizing block matrices of XTX∈Rk×k (where X∈Rn×k is a large data matrix obtained by high-fidelity simulations, the number of snapshot data is n≳109 rsim 10^9$$\\end{document}]]>, and the number of snapshots is k≲O(103)) without any approximations: for example, the singular value decomposition of X is replaced by the eigenvalue decomposition of XTX. Then, the computation of XTX is parallelized by utilizing the domain decomposition often used in flow simulations, which reduces the memory consumption for each parallel process and wall-clock time in the DMD by a factor approximately equal to the number of parallel processes. The parallel computation with communication is performed only for XTX, allowing for high parallel efficiency under massively parallel computations. Furthermore, the proposed exact parallel algorithm is extended to the Hankel DMD without any additional parallel computations, realizing the Hankel DMD of large-scale data collected by over a billion grid points with comparable cost and memory to the DMD without Hankel matrices. Moreover, this study shows that the Hankel DMD, which has been employed to enrich information and augment rank, is advantageous for large-scale high-dimensional data collected by high-fidelity simulations in data reconstruction and predictions of future states (while prior studies have reported such advantages for low-dimensional data). Several numerical experiments using large-scale data, including laminar and turbulent flows around a cylinder and transonic buffeting flow around a full aircraft configuration, demonstrate that (i) the proposed exact parallel algorithm reproduces the existing non-parallelized Hankel DMD, (ii) the Hankel DMD for large-scale data consisting of over a billion grid points is feasible by using the proposed exact parallel algorithm with high parallel efficiency on more than 6 thousand CPU cores, and (iii) the Hankel DMD has advantages for high-dimensional data such as n≳109 rsim 10^9$$\\end{document}]]>.Graphical abstract

CFD Code Parallelization on GPU and the Code Portability

AbstractGoal of this paper is to develop a fully functional parallel Computational Fluid Dynamics (CFD) code that is optimized to run on a single Graphics Processing Unit (GPU). This is achieved by writing the code in FORTRAN and OpenACC (Open Accelerators), providing them with an easily portable, platform independent code. Existing CFD code is significantly modified to allow for parallel asynchronous execution. Also, due to strong recursive dependencies in Tridiagonal Matrix Algorithm (TDMA) solver, it is replaced with Jacobi, which provides fast execution in environments with large number of parallel cores. In this research a computer code for simulation of 2D flow of water through the axisymmetric channel is used as a base for development. The parallel code is executed on GPU, single, and multicore Central Processing Unit (CPU), and the execution times are compared between platforms. Even though that Jacobi solver performs worse on single core computers, compared to its Gauss‐seidel counterpart, it is used to provide a baseline for comparison. In this work, it is shown that computation on finer grids takes less time on GPU than on CPU. The computation time increase with the number of cells in grid on GPU should follow the observed linear trend until the GPUs physical limitations are reached depending on memory size and core count.

Analysis and selection of the structure of a multiprocessor computing system according to the performance criterion

Objectives. Analysis of the various architectures of computing systems (CSs) used in recent decades has allowed us to identify the most common structures. One of the key features is the use of mass-produced equipment to create data processing subsystems (for example, multicore processors and high-capacity semiconductor memory), as well as network equipment to build communication subsystems. This reduces hardware costs and allows typical or cluster configurations to be created, which is especially important for expensive CSs. The desire to achieve high computational speed and performance in such CSs requires minimizing the time to complete the task and balancing time delays both in data processing subsystems and in the communication subsystem which provides data transmission inside the CS. The aim of this work is to analyze computing modules (CMs) and structures on the basis of which the construction of cluster CSs is carried out.Methods. The main results of the work were obtained using methods of mathematical analysis and modeling.Results. The study considers the structure of modern multicore microprocessors as the basis for building CMs of cluster CSs. As the number of cores in the microprocessor structure increases, the communication network which unites them into a single structure becomes more complicated. It has been shown that in new developments of microprocessors, communication between cores is performed in the form of a network. The microprocessors themselves are MIMD structures in accordance with the well-known Flynn classification.Conclusions. The proposed method of selecting an effective structure of a CS allows us to obtain the optimal structure of a CS according to the criterion of performance.

The Distinct Spatial Patterns and Physical Mechanisms of Coastal Boundary Layer Jets over the Northern South China Sea

Abstract Two distinct spatial patterns of coastal boundary layer jets in the northern South China Sea (CBLJ-NSCS) in June 2015–22 are investigated using hourly output data from the Weather Research and Forecasting Model with a horizontal grid spacing of 9 km and ERA5 data. Spatial pattern 1 shows a high-incidence core on the east side of Hainan Island, while pattern 2 features dual high-incidence cores covering both the east side of Hainan Island and the waters south of Guangdong, with the latter being stronger and more extensive. Unique diurnal cycles have been observed in different high-incidence cores of CBLJ-NSCS: the western core, on the east side of Hainan Island, peaks at night and shows a secondary afternoon subpeak, while the eastern core, located south of Guangdong, reaches its maximum intensity at night with a morning subpeak. Inertial oscillations triggered by large-scale perturbation wind circulation and momentum propagation from upstream Indochina Peninsula CBLJs explain the nocturnal enhancement of both cores. The thermal effect exerted by Hainan Island largely contributes to the afternoon enhancement in the western core. The formation of the morning subpeak of the eastern core results from strong convergence and lifting in the northeast region of it. When upstream CBLJs along the Annamite Range intensify, more momentum is propagated to the eastern core instead of the western one due to wind field changes, which promotes the transition from spatial pattern 1 to pattern 2. Significance Statement Two spatial patterns of coastal boundary layer jets in the northern South China Sea have been identified. Spatial pattern 1 shows a high-incidence core on the east side of Hainan Island, while pattern 2 features dual high-incidence cores covering both the east side of Hainan Island and the waters south of Guangdong. The transition between these two spatial patterns is related to low-level westerly winds along the coast of the Annamite Range. This study has uncovered the novel spatial pattern of coastal boundary layer jets in the northern South China Sea and provides new insights into precipitation research in southern China, considering the substantial influence of these jets on coastal weather.

Deep learning prediction of novel hollow core photonic crystal fiber with tuned As2S3 − LiNbO3 ring for multimode applications

Multimode capability in photonic crystal fiber (PCF) is an important feature in fiber development. This study presents an innovative Dual Ring Hollow core PCF (DRH-PCF), designed to support a large number of multimode operations with minimal loss and flat dispersion optimized in the wavelength range of 1.26 µm to 1.66 µm. The DRH-PCF structure comprises a high-purity silica matrix, featuring a unique configuration that includes a hollow core and high indexed Lithium niobate (LiNbO3)-Arsenic trisulfide (As2S3) dual rings of thickness 0.6 µm. This arrangement is complemented by a pattern of strategically placed air holes encircling the fiber. Our PCF is engineered to support an impressive 200 and more orbital angular momentum modes, with low confinement loss (∼10−5 dB/m) and near-flat dispersion (∼3 ps/mm/km) and very large power fraction (∼0.99), while maintaining high mode purity of >95 %. Furthermore, the fiber exhibits a near-flat dispersion profile over a wide spectral range, with a measured dispersion of ∼6 ps/(nm·km) around the operational wavelength. Stability due to fiber bend is also investigated for extreme bend tolerance (Bend radii RB=1cm to 80 cm). Deep Learning prediction is used to train and predict the performance of the fiber and obtained remarkably close results (within 5 %). This combination of features makes our PCF an excellent candidate for diverse applications in high-capacity optical communication systems, sensing technologies, and nonlinear optics.

ACCELERATING IMAGE PROCESSING USING PARALLEL COMPUTATIONS IN OPENMP: DEVELOPMENT AND PERFORMANCE ANALYSIS OF A GRAPHICS EDITOR

In today’s digital world, image processing plays a crucial role in various aspects of human life. From creating visual content for social media to analyzing medical images, powerful tools for image manipulation are in constant demand. The growing requirements for image quality and processing speed make the search for efficient methods to accelerate these processes highly relevant. This paper investigates the development and implementation of a graphics editor that utilizes OpenMP parallel computing to accelerate data processing. This technology enables the efficient distribution of computational tasks across multiple processor cores, significantly improving system performance. The developed software offers a set of popular graphic effects, including negative, grayscale, sepia, blur, sharpening, and posterization. Each effect is implemented in both sequential and parallel modes using OpenMP, allowing for the comparison of different computational approaches. To evaluate the effectiveness of the proposed solution, a series of experiments were conducted on images of various sizes. These experiments involved applying each effect to images ranging from small icons to high-quality photographs. A comparative analysis of the efficiency of sequential and parallel computation methods demonstrated the significant advantages of the latter. The results of the study show a substantial acceleration of image processing when using OpenMP technology. This acceleration is particularly noticeable for computationally intensive effects such as blurring or sharpening, and when working with large images. In some cases, it was possible to achieve a significant increase in processing speed, opening up new possibilities for working with large volumes of graphic data. This research has significant practical value for software developers working on optimizing the performance of graphics editors and other image processing applications. It demonstrates how the application of modern parallel computing technologies can significantly improve the efficiency of working with graphic data, paving the way for the creation of more powerful and faster image processing tools.

Generalized MPD and DMP Inverses

For a generalized Drazin invertible operator A and its core–EP inverse A ◯ ​​​​ d , the expression A A ◯ ​​​​ d A plays important role in the representations of well-known generalized inverses, in theory of partial orders and for decompositions of operator A. Solving certain system of operator equations leads to introduction of new generalized inverse of A A ◯ ​​​​ d A . Since this new inverse can be expressed by the generalized Moore–Penrose inverse and generalized Drazin inverse of A, it is called a generalized MPD inverse of A A ◯ ​​​​ d A . As a dual version of generalized MPD inverse, we also consider a generalized DMP inverse. The dual core inverse and core inverse of A, respectively, are special cases of the generalized MPD inverse and generalized DMP inverse. Several properties, representations and characterizations of generalized MPD and DMP inverses are established. Applying the generalized MPD and DMP inverses, we solve corresponding systems of linear equations and minimization problems. One well-known minimization problem is a particular case of our results.

Exhaustive search for novel multicomponent alloys with brute force and machine learning

We present an algorithm for the high-throughput computational discovery of intermetallic compounds in systems with a large number of components. It is particularly important for high entropy alloys (HEAs), where multiple principal elements can form numerous potential intermetallic compounds during the condensation process, making it challenging to predict the dominant phase. Our algorithm is based on a brute-force evaluation of candidate structures with a fixed underlying lattice (FCC or BCC) accelerated by machine-learning interatomic potentials. The algorithm takes a set of chemical elements and a crystal lattice type as inputs and produces structures on and near the convex hull of thermodynamically stable structures. The candidate structures are evaluated using the low-rank potential (LRP), trained to reproduce energies of structures equilibrated with density functional theory (DFT). Thanks to extreme computational effectiveness of the LRP, it is feasible to evaluate hundreds of thousands of structures per second, per CPU core. Thus, our algorithm screens a complete set of candidate structures for a given system without missing any configurations. We validated our method on systems with BCC (Nb-W, Nb-Mo-W, V-Nb-Mo-Ta-W) and FCC (Cu-Pt, Cu-Pd-Pt, Cu-Pd-Ag-Pt-Au) lattices and discovered 268 new alloys not reported in the AFLOW database1, which we used as a benchmark.

Workflow for high-quality visualisation of large-scale CFD simulations by volume rendering

High-fidelity CFD simulations can easily generate terabytes to petabytes of resulting data. Post-processing of such data is not an easy task. It holds especially for volume rendering, one of the most illustrative but computationally intensive post-processing techniques.This paper presents an HPC-ready workflow for post-processing large-scale CFD data computed on unstructured meshes by volume rendering using matured visual effects tools. The workflow consists of five steps: (1) parallel loading of unstructured data into memory, (2) data load-balancing among available resources, (3) re-sampling unstructured data into a regular grid (voxelisation), (4) storing data to OpenVDB format, and (5) final high-quality volume rendering of the (possibly sparse) regular grid in Blender. The workflow is based on open-source libraries, where we have improved all these steps to build an effective and robust approach. Due to parallel loading and appropriate load balancing, our workflow (a) allows loading sequential databases that do not fit into the memory of a single node and (b) significantly outperforms current scientific visualisation tools in voxelisation scalability. Moreover, due to the connection to professional visual effects tools such as Blender, interactive or photo-realistic volume rendering by path tracing, which includes global illumination effects, is allowed.With the workflow, it is possible to re-sample hundreds of time steps on an unstructured mesh with 1 billion cells (tens of TB of data) to a sparse regular grid with a density of 11 billion voxels and prepare data for interactive visualisation in just a few minutes using thousands of CPU cores.

AESware: Developing AES-enabled low-power multicore processors leveraging open RISC-V cores with a shared lightweight AES accelerator

As open-source RISC-V cores continue to be released, the development of low-power multicore processors utilizing these cores is invigorating the edge/IoT device market. Nevertheless, comprehensive research on developing low-power multicore processors with integrated security features using existing open RISC-V cores remains limited. This study addresses this gap by introducing AESware, a dedicated lightweight hardware designed for energy-efficient AES (Advanced Encryption Standard) task execution, contributing to the development of AES-specific low-power RISC-V multicore processors. AESware supports variable key lengths and ensures minimal power consumption with a compact design. This standalone IP (Intellectual Property) is compatible with various open RISC-V cores, offering scalability and convenience. And importantly, we propose the most energy-efficient architecture for multicore processors equipped with AESware. Instead of assigning dedicated AESware to each core, we introduce a shared AESware architecture to maximize energy efficiency. We develop an operational algorithm for task scheduling in AESware, achieving maximum utilization and minimal latency while maintaining its lightweight nature. To evaluate our solution, we developed 24 processors into three groups: AESware-equipped, baseline, and those with an external AES accelerator per core. After FPGA (Field-Programmable Gate Array) prototyping for functional verification and power consumption analysis via 45 nm process technology synthesis, our findings revealed significant energy savings. AESware-equipped processors achieved up to 76%, 47%, and 33% energy savings at dual-, quad-, and octa-core configurations compared to baseline, respectively, and were more energy-efficient in running AES applications than those with individual accelerators.

DecTree: a physics-based geochemical surrogate for surface complexation of uranium on clay

Abstract. Geochemistry is usually the computational bottleneck in coupled reactive transport simulations, which hampers the complexity of the systems and of the processes they can investigate. In recent years, promising speedups have been obtained by substituting the numerical solution of geochemical models with approximated surrogates borrowed from artificial intelligence and machine learning (AI/ML). In the framework of the DONUT/EURAD project a set of benchmarks were defined to assess the performance and the accuracy of different surrogate approaches in settings relevant to the safety assessment of nuclear waste repositories, such as the surface complexation and exchange of U(VI) on clay. In this context, this work introduces am original surrogate modelling approach based on recursive partitioning of parameter space, which exploits prior domain knowledge for the training. The surrogate, which can be represented as a decision tree, hence the DecTree name, performs dimensionality reduction by identifying functional relationships between outputs and input variables using a straightforward non-monotonic extension of the Spearman's rank correlation coefficient. New predictions are then interpolated from the partitioned training data. Applied to a low-dimensional geochemical model, DecTree shows virtually no training time and excellent accuracy, ensuring a throughput of around 500 000 predictions per second on a single CPU core.

An assessment of large language models for OpenMP-based code parallelization: a user perspective

Large language models have sparked a lot of attention in the research community in recent days, especially with the introduction of practical tools such as ChatGPT and Github Copilot. Their ability to solve complex programming tasks was also shown in several studies and commercial solutions increasing the interest in using them for software development in different fields. High performance computing is one of such fields, where parallel programming techniques have been extensively used to utilize raw computing power available in contemporary multicore and manycore processors. In this paper, we perform an evaluation of the ChatGPT and Github Copilot tools for OpenMP-based code parallelization using a proposed methodology. We used nine different benchmark applications which represent typical parallel programming workloads and compared their OpenMP-based parallel solutions produced manually and using ChatGPT and Github Copilot in terms of obtained speedup, applied optimizations, and quality of the solution. ChatGPT 3.5 and Github Copilot installed with Visual Studio Code 1.88 were used. We concluded that both tools can produce correct parallel code in most cases. However, performance-wise, ChatGPT can match manually produced and optimized parallel code only in simpler cases, as it lacks a deeper understanding of the code and the context. The results are much better with Github Copilot, where much less effort is needed to obtain correct and performant parallel solution.

Embankment Breach Simulation and Inundation Mapping Leveraging High-Performance Computing for Enhanced Flood Risk Prediction and Assessment

Abstract. Embankment breaches represent significant hazards to communities and infrastructure, precipitating catastrophic flood occurrences. The precise prediction of floods and understanding the scope of inundation stemming from embankment failures are imperative for effective disaster preparedness and response. This research delves into a case study on simulating embankment breaches to evaluate the extent of flooding. Leveraging advanced hydrodynamic models validated through high-performance computing (HPC) systems, and integrating real-time data assimilation, we aim to improve accuracy in flood forecasting. The study endeavours to bolster flood risk management by furnishing detailed inundation maps and insights into embankment breach dynamics, thereby facilitating enhanced preparedness and response strategies. Our findings reveal that simulations conducted on multicore processors offer superior performance compared to single-core setups, yielding enhanced result accuracy and providing administrators with increased lead time. It unlocks high-resolution simulations for intricate basin details, explores a wider range of flood scenarios quickly, and allows for efficient ensemble modelling to assess model uncertainty. Through HPC utilization, we can harness high-resolution digital elevation models (DEMs) for 2D-hydrodynamic modelling, enabling rapid assessment of water spread resulting from embankment breaches within a mere 20-minute timeframe, a significant improvement from the previous 3–4 hours duration.

Innovative techniques for developing hybrid yarns using different core insertion methods: A comparative study

Elastane fibers, in particular, have gained an important share in the production of elastic yarn. This research aims to evaluate innovative methods for developing hybrid yarns (dual core) by integrating core elements in various configurations to show the influence of different core-inserting techniques on yarn characteristics. The research developed a modified pulley (W-grooved roller) to insert core elements side by side along with the existing method during the yarn manufacturing process. Ne 16/1 yarn (dual core) was manufactured using five different core inserting techniques with cotton roving, polybutylene terephthalate (PBT), and elastane (Lycra®). The characteristics of the produced yarns were analyzed using standard testing methods. Statistical analysis was performed to determine significant variations in yarn properties using one-way MANOVA and the MOORA method was employed to determine the most suitable yarn configuration. The results indicate that yarn manufactured by inserting core components side by side significantly enhances the expected characteristics of hybrid yarns. Yarns manufactured by inserting core components side by side show higher strength, elongation, and lower hairiness values. This research introduces a novel technique for producing dual-core yarns by side-by-side insertion of core elements contributing to the development of innovative hybrid yarns.

A Parallel Tag Cache for Hardware Managed Tagged Memory in Multicore Processors

A Parallel Tag Cache for Hardware Managed Tagged Memory in Multicore Processors

Experimental study of a distributed active noise control system with multi-device nodes based on augmented diffusion strategy.

Recently, distributed active noise control (DANC) algorithms have been explored as a way to reduce computational complexity while ensuring system stability, thereby outperforming conventional centralized and decentralized algorithms. Most existing DANC algorithms assume that each node has only one pair of loudspeaker and microphone, limiting their flexibility in practical applications. In contrast, this paper proposes a DANC algorithm with general multi-device nodes based on the recently developed augmented diffusion strategy, allowing flexible and scalable ANC applications. A real-time distributed ANC system based on a multi-core digital signal processor platform is developed in order to compare the control performance of the proposed extended augmented diffusion algorithm with that of existing centralized, decentralized and augmented diffusion algorithms. Real-time experiments demonstrate that the proposed algorithm exhibits noise reduction performance consistent with that of the centralized algorithm while achieving lower global computational complexity and avoiding the system instability risk of the decentralized algorithm. Further, the new algorithm improves convergence speed and reduces the global communication cost compared to the previous augmented diffusion algorithm. Experimental results indicate the application potential of the proposed DANC algorithm for a generalized system configuration.

Improved automated parallel implementation of GMM background subtraction on a multicore digital signal processor

<p><span>Scene segmentation is an essential step in a wide range of video processing applications, for instance, object recognition and tracking. The Gaussian mixture model (GMM) for background subtraction (BS) has gained widespread usage in scene segmentation, despite its known computational intensity. To tackle this challenge, we propose a practical solution to accelerate processing through a parallel implementation on an embedded multicore platform. In this paper, we present an improved automated parallel implementation of the GMM algorithm using the Orphan directive provided by open multiprocessing (OpenMP). Experimental assessments conducted on the eight cores of the C6678 digital signal processor (DSP) demonstrate significant advancements in parallel efficiency, particularly when handling high-resolution frames, including high-definition (HD) and full-HD resolutions. The achieved parallel efficiency surpasses the results obtained with classical OpenMP scheduling modes, encompassing dynamic, static, and guided approaches. Specifically, the parallel efficiency reaches approximately 82% for full-HD resolution frames and, 99.3% for low-resolution frames, respectively.</span></p>

Research on the application of load balancing strategy in IEC61850 communication of large-scale new energy monitoring system

Abstract This paper proposes an optimization scheme to achieve high throughput and stability of the system through efficient thread management and load-balancing strategies. Firstly, based on the IEC 61850 standard framework, a distributed architecture is constructed, which uses the advantages of multi-core processors for task distribution to improve data processing capabilities and system response speed. Secondly, in terms of thread management, prioritizing key operations ensures the timely processing of important data and reduces latency. At the same time, an asynchronous processing mechanism is introduced, so that the control and data can work in parallel, avoiding the occurrence of blockage. Then, the load-balancing strategy reasonably distributes the tasks to different processing nodes, avoiding the problem of single-point overload. Finally, the data caching mechanism and high-speed data bus are specially strengthened in the design to support the efficient transmission of large amounts of real-time data.

Optimal capacity planning for cloud service providers with periodic, time-varying demand

Allocating capacity to private cloud computing services is challenging because demand is time-varying, there are often no buffers, and customers can re-submit jobs a finite number of times. We model this setting using a multi-station queueing network where servers represent CPU cores and jobs not immediately processed retry several times. Assuming retrial rates are stationary and that there is a maximum number of retrial attempts, we determine an optimal service capacity and retrial interval under an admission control policy employed by our partner institution — the server informs customers when they should next attempt service without enforcement. We introduce a recursive representation of the offered load which approximates the fluid dynamics of the system. We then use this representation to develop a solution technique that minimizes the total variation in the constructed offered load. We prove this approach is linked to maximizing system throughput and that in certain settings, the optimal stationary and time-varying retrial intervals are equivalent. Utilizing a data set of cloud computing requests spanning a 24-hour period, our analysis indicates that the optimal policy prescribes a 10% reduction in capacity. We also investigate the fidelity of the fluid model and the sensitivity of our recommendations to the behavior of retrial jobs. We find that retrial-time announcements allow a provider to satisfy service level agreements while encouraging retrial jobs to be processed during off-peak periods. Further, the policy is suitably robust to a customer’s willingness to comply with the suggested retrial times.