Parallel Tasks Research Articles

Performance Optimization in Three-Modality Biometric Verification using Heterogeneous CPU-GPU Computation

This paper proposes a method to improve the performance of tri-modal biometric verification using a heterogeneous computing system exploiting the synergy between CPU and GPU. The main objective is to reduce the time required for verification while maintaining the system's accuracy. The design of this system is based on a decision fusion algorithm based on the logical OR connector, enabling the results of the three modalities to be combined. The implementation is being carried out in C# with Visual Studio 2019, using the Task Parallel Library to parallelize tasks on the CPU, and OpenCL.NET to manage processing on the GPU. The tests carried out on a representative sample of 1,000 individuals, show a clear improvement in performance compared with a sequential system. Execution times were significantly reduced, ranging from 0.03 ms to 0.67 ms for data sizes between 50 and 1000. Analysis of the performance gains, based on Amdahl's law, reveals that the proportion of tasks that can be parallelized remains higher in heterogeneous systems than in parallel and sequential systems, even though part of processing remains sequential for large data sizes. This study highlights the ability of heterogeneous computing systems to effectively reduce the verification time of biometric systems while maintaining an optimal balance between processing speed and overall efficiency. The results demonstrate the potential of this approach for advanced biometric applications, particularly in distributed environments.

The impact of distributed systems on the architecture and design of computer systems: advantages and challenges

A distributed system can encompass a variety of configurations, including mainframes, personal computers, workstations, and minicomputers. The varying degrees of software flexibility and the ability to execute tasks in parallel facilitate simultaneous data processing across multiple processors. The higher the resilience of an application, the quicker it can recover after a system failure. Organisations increasingly adopt distributed computing systems as they face increased data generation and demand for enhanced application performance. These systems enable businesses to scale effectively in response to growing data volumes. Integrating additional hardware into a distributed system is generally simpler than upgrading a centralised system reliant on powerful servers. Distributed systems comprise numerous nodes that collaborate towards a common objective. This article aims to provide a comprehensive overview of distributed systems, their architectural frameworks, and essential components. This study examines how distributed systems influence the architecture and design of computer systems. The research methods consist of reviewing existing literature and analysing case studies on implementing distributed systems. Key findings indicate that the evolution of distributed systems is ongoing, driven by emerging technologies and the increasing demand for efficient, scalable, and secure solutions. Innovations such as edge computing, blockchain technology, 5G, and the integration of AI and machine learning are among the notable trends shaping the future landscape of distributed systems. Looking ahead, designers and architects need to stay informed about these advancements to create reliable and adaptable distributed systems that can address the dynamic needs of users and organisations

USING ASYNCHRONOUS PROGRAMMING IN PYTHON TO IMPROVE APPLICATION PERFORMANCE

Asynchronous programming in Python is a powerful tool for improving application performance by effectively managing multitasking. The main focus is on the use of the async and await keywords, as well as the asyncio, ThreadPoolExecutor and ProcessPoolExecutor libraries, which play an important role in organizing multitasking processes. This is especially true for applications related to I/O operations, such as web servers and APIs. The use of asynchronous programming allows you to eliminate thread locks and optimize query processing, which makes applications more responsive and scalable. The article provides specific examples of using the asynchronous approach in practice, including parallel task execution and resource management. In addition, the study demonstrates that the introduction of asynchronous technologies helps to reduce infrastructure costs, ensuring high throughput and stable operation even under high load conditions. Asynchronous programming stands out for its flexibility and the ability to create high-performance systems that are capable of handling a large number of simultaneous I/O operations.

Coarse-Grained Task Parallelization by Dynamic Profiling for Heterogeneous SoC-Based Embedded System

In this study, we introduce a methodology for automatically transforming user applications written in C/C++ to a parallel representation consisting of coarse-grained tasks based on dynamic profiling. Such a parallel representation is suitable for mapping applications onto heterogeneous SoCs. We present our approach for instrumenting the user application binary during the compilation process with parallel primitives that enable the runtime system to schedule and execute independent computation-intensive coarse-grained tasks concurrently. We use the proposed compilation and code transformation methodology to retarget each application for execution on a heterogeneous SoC composed of processor cores and accelerators. We demonstrate the capabilities of our integrated compile time and runtime flow through task-level parallelization and functionally correct execution of real-world applications in the communication systems and radar processing domains. We demonstrate the functionality of our integrated system by executing six distinct applications with different degrees of parallelism on four different platforms: an eight-core general-purpose processor, a heterogeneous SoC simulator, and two heterogeneous SoCs utilizing the Xilinx Zynq UltraScale+ FPGA and the Nvidia Jetson AGX board. Our integrated approach offers a path forward for application developers to take full advantage of the target SoC without requiring users to become hardware or parallel programming experts.

Energy-harvesting-aware federated scheduling of parallel real-time tasks

Energy-harvesting-aware federated scheduling of parallel real-time tasks

SARS-CoV-2 Illumina GeNome Assembly Line (SIGNAL), a Snakemate workflow for rapid and bulk analysis of Illumina sequencing of SARS-CoV-2 genomes.

The incorporation of sequencing technologies in frontline and public health healthcare settings was vital in developing virus surveillance programs during the Coronavirus Disease 2019 (COVID-19) pandemic caused by transmission of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, increased data acquisition poses challenges for both rapid and accurate analyses. To overcome these hurdles, we developed the SARS-CoV-2 Illumina GeNome Assembly Line (SIGNAL) for quick bulk analyses of Illumina short-read sequencing data. SIGNAL is a Snakemake workflow that seamlessly manages parallel tasks to process large volumes of sequencing data. A series of outputs are generated, including consensus genomes, variant calls, lineage assessments and identified variants of concern (VOCs). Compared to other existing SARS-CoV-2 sequencing workflows, SIGNAL is one of the fastest-performing analysis tools while maintaining high accuracy. The source code is publicly available (github.com/jaleezyy/covid-19-signal) and is optimized to run on various systems, with software compatibility and resource management all handled within the workflow. Overall, SIGNAL illustrated its capacity for high-volume analyses through several contributions to publicly funded government public health surveillance programs and can be a valuable tool for continuing SARS-CoV-2 Illumina sequencing efforts and will inform the development of similar strategies for rapid viral sequence assessment.

Enhanced photonic reservoir computing using an optically injected VCSEL with random polarized optical feedback.

We propose and numerically demonstrate a photonic time-delay reservoir computing (TDRC) system exhibiting enhanced parallel task processing performance, where an optically injected vertical-cavity surface-emitting laser (VCSEL) under random distributed optical feedback acts as the reservoir computer. To assess its effectiveness, we perform two benchmark tasks including chaotic time-series prediction and waveform recognition task, where the TDRC is associated with two different random feedback structures, i.e., orthogonally polarized optical feedback (OPOF) and parallelly polarized optical feedback (PPOF). Benefiting from the enhanced nonlinearity offered by the random distributed optical feedback, the proposed TDRC excels at parallel task processing with the PPOF structure, whereas the performance of the OPOF structure may be deteriorated. Additionally, we reveal the effect of the injection strength, feedback strength, pump current, and number of virtual nodes on the proposed TDRC. Our work paves the way for the performance enhancement of parallel task processing based on polarization multiplexing in a VCSEL-based TDRC.

A Task Parallel Accelerator with Dynamic Pipeline Balancing

Coarse-grained reconfigurable array(CGRA) based ac- celerator is a promising architecture to accelerate data processing workloads. CGRA-based accelerator features more flexibility in adopting various workloads while remaining powerful in comparison to the traditional application specified accelerator. However, the strength of CGRAs is limited by irregular data access and dependence patterns. The previously proposed task- based execution model enabled work-aware dynamic scheduling, but the effectiveness is still limited by the regular execution resources. To address these issues, we proposed a heterogeneous architecture to improve parallelism for pipeline-enabled task streaming. We enabled dynamic pipeline balancing on this architecture with minor modifications to the task stream annotation. We compare the execution result on various heterogeneous structures with the regular configuration. Overall, we find that our architecture can improve performance with the same overall resources.

Online level-2 perspective taking for newly learnt symbols.

Humans demonstrate spontaneous sensitivity to other people's perspectives on object identities in online tasks. Evidence shows that this not only involves representing the mere discrepancy between perspectives, but the content of such perspectives as well (level-2 perspective taking/L2PT). However, this evidence comes from studies using culturally grounded symbols which leaves open the possibility that having extensive, easily accessible background knowledge about an object is necessary for the L2PT effect. Experiment 1 tested this by comparing L2PT across two groups: one performing a verification task on Arabic numbers, and one on newly learnt symbol-label pairs. In both groups, half of the visual stimuli was symmetrical, while half was asymmetrical. In both cases, there was a joint condition: participants performed the task in parallel with a partner, observing stimuli from opposite angles, thus having conflicting interpretations for asymmetric characters. Furthermore, they also performed the verification task individually, while their partner had no visual access to the stimuli. We found an interference effect in both groups. However, while the effect was stable in the number group, it diminished over time in the symbol group. Experiments 2a and 2b demonstrated that the complexity of the recently learnt symbols has an influence on spontaneous L2PT: the same procedure with more complex symbols did not elicit any interference effect. Our results show that online L2PT is not limited to objects that participants have proficiency in identifying. Nevertheless, the L2PT effect seems to diminish when participants have to process increasingly complex novel symbols.

ExaGRyPE: Numerical general relativity solvers based upon the hyperbolic PDEs solver engine ExaHyPE

ExaGRyPE describes a suite of solvers and solver ingredients for numerical relativity that are based upon ExaHyPE 2, the second generation of our Exascale Hyperbolic PDE Engine. Numerical relativity simulations are crucial in resolving astrophysical phenomena in strong gravitational fields and are fundamental in analyzing and understanding gravitational wave emissions. The presented generation of ExaGRyPE solves the Einstein field equations in the standard CCZ4 formulation under a 3+1 foliation and focuses on black hole space-times. It employs a block-structured Cartesian grid carrying a higher-order Finite Difference scheme with full support of adaptive mesh refinement (AMR), it facilitates massive parallelism combining message passing, domain decomposition and task parallelism, and it supports the injection of particles into the grid as static data probes or as moving tracers. We introduce the ExaGRyPE-specific building blocks within ExaHyPE 2, and discuss its software architecture and compute-n-feel.For this, we formalize the creation of any specific astrophysical simulation with ExaGRyPE as a sequence of lowering operations, where a few abstract logical tasks are successively broken down into finer and finer tasks until we obtain an abstraction level which can directly be mapped onto a C++ executable. The overall program logic is fully specified via a domain-specific Python interface, we automatically map this logic onto a more detailed set of numerical tasks, subsequently lower this representation onto technical tasks that the underlying ExaHyPE engine uses to parallelize the application, before eventually the technical tasks in turn are mapped onto small task graphs including the actual astrophysical PDE term evaluations, initial conditions, boundary conditions, and so forth. These can be injected manually by the user, or users might instruct the solver on the most abstract user interface level to use out-of-the-box ExaGRyPE implementations. We end up with a rigorous separation of concerns which shields ExaGRyPE users from technical details and hence simplifies the development of novel physical models. We present the simulations and data for the gauge wave, static single black holes and rotating binary black hole systems, demonstrating that the code base is mature and usable. However, we also uncover domain-specific numerical challenges that need further study by the community in future work.

Adaptive memory reservation strategy for heavy workloads in the Spark environment.

The rise of the Internet of Things (IoT) and Industry 2.0 has spurred a growing need for extensive data computing, and Spark emerged as a promising Big Data platform, attributed to its distributed in-memory computing capabilities. However, practical heavy workloads often lead to memory bottleneck issues in the Spark platform. This results in resilient distributed datasets (RDD) eviction and, in extreme cases, violent memory contentions, causing a significant degradation in Spark computational efficiency. To tackle this issue, we propose an adaptive memory reservation (AMR) strategy in this article, specifically designed for heavy workloads in the Spark environment. Specifically, we model optimal task parallelism by minimizing the disparity between the number of tasks completed without blocking and the number completed in regular rounds. Optimal memory for task parallelism is determined to establish an efficient execution memory space for computational parallelism. Subsequently, through adaptive execution memory reservation and dynamic adjustments, such as compression or expansion based on task progress, the strategy ensures dynamic task parallelism in the Spark parallel computing process. Considering the cost of RDD cache location and real-time memory space usage, we select suitable storage locations for different RDD types to alleviate execution memory pressure. Finally, we conduct extensive laboratory experiments to validate the effectiveness of AMR. Results indicate that, compared to existing memory management solutions, AMR reduces the execution time by approximately 46.8%.

Bimodal aphasia and dysgraphia: Phonological output buffer aphasia and orthographic output buffer dysgraphia in spoken and sign language

We report a case of crossmodal bilingual aphasia—aphasia in two modalities, spoken and sign language—and dysgraphia in both writing and fingerspelling. The patient, Sunny, was a 42 year-old woman, after a left temporo-parietal stroke, a speaker of Hebrew, Romanian, and English and an adult learner, daily user of Israeli Sign language (ISL).We assessed Sunny's spoken and sign languages using a comprehensive test battery of naming, reading, and repetition tasks, and also analysed her spontaneous-speech and sign. Her writing and fingerspelling were assessed using tasks of dictation, naming, and delayed copying.In spoken language production, Sunny showed a classical phonological output buffer (POB) impairment in naming, reading, repetition, and spontaneous production, with phonological errors (transpositions, substitutions, insertions, and omissions) in words and pseudo-words, and whole-unit errors in morphological affixes, function-words, and number-words, with a length effect. Importantly, her error pattern in ISL was remarkably similar in the parallel tasks, with phonological errors in signs and pseudo-signs, affecting all the phonological parameters of the sign (movement, handshape, location, and orientation), and whole-unit errors in morphemes, function-signs, and number-signs. Sunny's impairment was selective to the POB, without phonological input, semantic-conceptual, or syntactic deficits. This shows for the first time how POB impairment, a kind of conduction aphasia, manifests itself in a sign language, and indicates that the POB for sign-language has the same cognitive architecture as the one for spoken language. It may also indicate similar neural underpinnings for spoken and sign languages.In writing, Sunny forms the first case of a selective type of dysgraphia in fingerspelling, orthographic (graphemic) output buffer dysgraphia. In both writing and fingerspelling, she made letter errors (letter transpositions, substitutions, insertions, and omissions), as well as morphological errors and errors in function words, and showed length effect. Sunny's impairment was selective to the orthographic output buffer, whereas her reading, including orthographic input processing, was intact. This suggests that the orthographic output buffer is shared for writing and in fingerspelling, at least in a late learner of sign language. The results shed further light on the architecture of phonological and orthographic production.

EdgeOPT: A Competitive Algorithm for Online Parallel Task Scheduling With Latency Guarantee in Mobile Edge Computing

EdgeOPT: A Competitive Algorithm for Online Parallel Task Scheduling With Latency Guarantee in Mobile Edge Computing

SPSC: Stream Processing Framework Atop Serverless Computing for Industrial Big Data.

With the advance of smart manufacturing and information technologies, the volume of data to process is increasing accordingly. Current solutions for big data processing resort to distributed stream processing systems, such as Apache Flink and Spark. However, such frameworks face challenges of resource underutilization and high latency in big data application scenarios. In this article, we propose SPSC, a serverless-based stream computing framework where events are discretized into the atomic stream and stateless Lambda functions are taken as context-irrelevant operators, achieving task parallelism and inherent data parallelism in processing. Also, we implement a prototype of the framework on Amazon Web service (AWS) using AWS Lambda, AWS simple queue service, and AWS DynamoDB. The evaluation shows that compared with Alibaba's real-time computing Flink version, SPSC outperforms by 10.12% when the overhead is close.

There Is More to Avatars Than Visuals: Investigating Combinations of Visual and Auditory User Representations for Remote Collaboration in Augmented Reality

Supporting remote collaboration through augmented reality facilitates the experience of co-presence by presenting the collaborator’s avatar in the user’s physical environment. While visual user representations are continuously researched and advanced, the audio configuration – especially in combination with different visualizations – is rarely considered. In a user study (n = 48, 24 dyads), we evaluate the combination of two visual (Simple vs. Rich Avatar) with two auditory (Mono vs. Spatial Audio) user representations, to investigate their impact on user’s overall experience, performance, and perceived social presence during collaboration. Our results show a preference for rich auditory and visual user representations, as Spatial Audio supports completion of parallel tasks and the Rich Avatar positively influence user experience. However, the Simple Avatar draws less attention, which potentially benefits task efficiency, advocating for simpler visualizations in performance-oriented settings. Our findings contribute to a deeper understanding of how visual and auditory user representation combinations impact remote collaboration in augmented reality.

Photonic reservoir computing for parallel task processing based on a feedback-free spin-polarized VCSEL

Time delayed reservoir computing (RC) is a novel artificial neural network that is easy to implement in hardware due to its extremely simple structure. Because of its time-division multiplexed information processing, laser-based photonic time-delayed RCs usually realize parallel processing with polarization/wavelength multiplexing. However, the performance of two different tasks is difficult to regulate separately and simultaneously in the time delayed RC system, especially for the chip-scale configuration. Here, we propose a feedback-free RC system based on a spin-polarized vertical-cavity surface-emitting semiconductor laser (VCSEL), which simplifies the whole system structure and can process time series prediction and waveform recognition tasks in parallel, with employing the input and output coding to provide the effect from past states. By separately setting the number of past states introduced by the coding for the two tasks, the performance of the two tasks can be adjusted respectively. Furthermore, by appropriately tuning the pump polarization ellipticity which is the unique feature for the spin-polarized VCSEL, the computational ability of the proposed RC can be focused on one of the two parallel tasks. Therefore, the proposed RC system is capable of dealing with different tasks with high performance, and also expected to provide a viable solution for integrated neuromorphic computing systems due to its compact, feedback-free structure.

Research on Parallel Task Scheduling Algorithm of SaaS Platform Based on Dynamic Adaptive Particle Swarm Optimization in Cloud Service Environment

To efficiently realize the parallel task scheduling of SaaS platform in large-scale cloud service environment, this paper studies the parallel task scheduling algorithm of SaaS platform based on dynamic adaptive particle swarm optimization in cloud service environment. Users access the cloud through the user access interface module, and issue task scheduling instructions or send task scheduling requests. After the service management module provides diversified application service support according to the scheduling requirements, the core service module determines the SaaS platform parallel scheduling objective function, and uses dynamic adaptive particle swarm optimization to solve the objective function to obtain the SaaS platform parallel task scheduling results. The test results show that the algorithm has better multi-objective solving ability and can obtain higher quality objective solutions, and the test results of the total execution time of parallel scheduling tasks and the total transmission time of task data on SaaS platform are all within 30 s. The results of virtual machine resource load balancing degree are all below 15%; the utilization rate of virtual machine resources is above 92.2%.

Artificial potential field-empowered dynamic holographic optical tweezers for particle-array assembly and transformation

Owing to the ability to parallel manipulate micro-objects, dynamic holographic optical tweezers (HOTs) are widely used for assembly and patterning of particles or cells. However, for simultaneous control of large-scale targets, potential collisions could lead to defects in the formed patterns. Herein we introduce the artificial potential field (APF) to develop dynamic HOTs that enable collision-avoidance micro-manipulation. By eliminating collision risks among particles, this method can maximize the degree of parallelism in multi-particle transport, and it permits the implementation of the Hungarian algorithm for matching the particles with their target sites in a minimal pathway. In proof-of-concept experiments, we employ APF-empowered dynamic HOTs to achieve direct assembly of a defect-free 8 × 8 array of microbeads, which starts from random initial positions. We further demonstrate successive flexible transformations of a 7 × 7 microbead array, by regulating its tilt angle and inter-particle spacing distances with a minimalist path. We anticipate that the proposed method will become a versatile tool to open up new possibilities for parallel optical micromanipulation tasks in a variety of fields.

Better together: Automated app review analysis with deep multi-task learning

Context:User reviews of mobile apps provide an important communication channel between developers and users. Existing approaches to automated app review analysis mainly focus on one task (e.g., bug classification task, information extraction task, etc.) at a time, and are often constrained by the manually defined patterns and the ignorance of the correlations among the tasks. Recently, multi-task learning (MTL) has been successfully applied in many scenarios, with the potential to address the limitations associated with app review mining tasks. Objective:In this paper, we propose MABLE, a deep MTL-based and semantic-aware approach, to improve app review analysis by exploiting task correlations. Methods:MABLE jointly identifies the types of involved bugs reported in the review and extracts the fine-grained features where bugs might occur. It consists of three main phases: (1) data preparation phase, which prepares data to allow data sharing beyond single task learning; (2) model construction phase, which employs a BERT model as the shared representation layer to capture the semantic meanings of reviews, and task-specific layers to model two tasks in parallel; (3) model training phase, which enables eavesdropping by shared loss function between the two related tasks. Results:Evaluation results on six apps show that MABLE outperforms ten commonly-used and state-of-the-art baselines, with the precision of 79.76% and the recall of 79.24% for classifying bugs, and the precision of 79.83% and the recall of 80.33% for extracting problematic app features. The MTL mechanism improves the F-measure of two tasks by 3.80% and 4.63%, respectively. Conclusion:The proposed approach provides a novel and effective way to jointly learn two related review analysis tasks, and sheds light on exploring other review mining tasks.

Optimization of office process task allocation based on deep reinforcement learning

In office platforms, we often need to face the situation of a large number of parallel heterogeneous process tasks. This not only tests the ability of the task executors themselves, but also puts forward requirements for the performance of the collaborative scheduling system. Using the reinforcement learning method, combined with quantitative analysis such as collaborative coordination and slackness, and based on the Markov Bollinger theory, a multi-agent Bollinger model is proposed to realize the optimization scheduling system with overall process coordination and maximum completion time as the optimization objectives, thereby improving the overall execution efficiency. Taking the real business system process as the test scenario, under the same optimization objective, the reinforcement learning algorithm based on D3QN and DRL and the meta-heuristic algorithm based on ant colony are compared to verify the effectiveness of the proposed method.