检索结果-内蒙古大学图书馆

international conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)

作者： Nelson Lungu Bibhuti Bhusan Dash Utpal De Chandra Bibek Bikram Dash Binod Kumar Pattanayak Sudhansu Shekhar Patra Electrical and Electronical Engineering University of Zambia Lusaka Zambia School of Computer Applications KIIT Deemed to be University Bhubaneswar India HealthWorksAI Bangalore India Department of Computer Science & Engineering ITER Siksha ‘O’ Anusandhan University Bhubaneswar Odisha India

ISBN: (数字)9798350376425

ISBN: (纸本)9798350376432

Graphics processing units (GPUs) have become ubiquitous in computing systems, from personal devices to data centres. Their massively parallel architecture delivers tremendous performance for graphics and general-purpose workloads. However, the open nature of GPUs also introduces security vulnerabilities that can be exploited through side-channel attacks. Defending against such threats often incurs performance overhead, creating tradeoffs between security and speed. This article examines recent research analysing and mitigating GPU side channel attacks while maintaining acceptable performance and presented a multi-faceted defence strategy combining compiler transformations, runtime monitoring, and microarchitectural techniques to balance security and performance. The methodology integrates formal threat models, empirical side-channel evaluations, and rigorous benchmarking. It implements and evaluates prototypes demonstrating security improvement with minimal impact on representative workloads. The results provide promising directions for practical GPU security defences.

关键词： Performance evaluation Threat modeling Runtime Microarchitecture Prevention and mitigation Graphics processing units Prototypes Side-channel attacks Security Protection

来源：评论

学校读者我要写书评

暂无评论

Performance and Metrics Analysis Between Python3 via Mojo

Performance and Metrics Analysis Between Python3 via Mojo

引用

Sustainable Computing and Smart Systems (ICSCSS), international conference on

作者： Anuj Kumar Aditya Deo Swayam Gupta Roumo Kundu Piyush Jaiswal Taha Fatma Mohan Kumar Dehury Amity Institute of Information Technology Amity University Jharkhand Ranchi India

ISBN: (数字)9798350379990

ISBN: (纸本)9798350391558

In the field of programming languages, Mojo and Python have gained significant popularity and recognition among developers. While Mojo, a newly emerging language, and Python, a well-established language, both have their own merits and features that make them suitable for various programming tasks. Recent AI techniques such as transformers in Natural Language processing (NLP), Reinforcement Learning (RL), and Generative Adversarial Networks (GANs) have shown remarkable advancements. However, these techniques face challenges like high computational costs, scalability issues, and integration complexity. While Python features user-friendly syntax and extensive libraries, its interpreted nature can hinder performance for computationally intensive tasks. This research addresses this limitation by introducing Mojo, a high-performance language specifically designed for AI applications. Mojo leverages compilation and advanced optimization techniques to achieve significantly faster execution speeds compared to Python. The Mojo programming language can address these challenges by offering high-performance computation, efficient memory management, and seamless integration with AI frameworks. This can lead to faster processing times, better scalability, and more streamlined development workflows for advanced AI systems. The paper presents empirical evidence demonstrating the substantial performance gains offered by Mojo. Furthermore, the analysis explores several advantages of Mojo beyond raw speed. These include static typing, which enhances code reliability and maintainability, and built-in support for parallelism, enabling efficient utilization of multi-core processors. Additionally, Mojo's seamless integration with existing Python codebases allows developers to leverage the extensive Python ecosystem while enjoying the performance benefits of Mojo.

关键词： Scalability Syntactics parallel processing Natural language processing Libraries Computational efficiency Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

27th international European conference on parallel and distributed Computing, Euro-Par 2021

27th International European Conference on Parallel and Distr...

引用

27th international European conference on parallel and distributed Computing, Euro-Par 2021

ISBN: (纸本)9783030856649

The proceedings contain 38 papers. The special focus in this conference is on parallel and distributed Computing. The topics include: An MPI-based Algorithm for Mapping Complex Networks onto Hierarchical Architectures;pipelined Model parallelism: Complexity Results and Memory Considerations;efficient and Systematic Partitioning of Large and Deep Neural Networks for parallelization;A GPU Architecture Aware Fine-Grain Pruning Technique for Deep Neural Networks;Towards Flexible and Compiler-Friendly Layer Fusion for CNNs on Multicore CPUs;smart distributed DataSets for Stream processing;colony: parallel Functions as a Service on the Cloud-Edge Continuum;horizontal Scaling in Cloud Using Contextual Bandits;geo-distribute Cloud applications at the Edge;Automatic Low-Overhead Load-Imbalance Detection in MPI applications;A Fault Tolerant and Deadline Constrained Sequence Alignment Application on Cloud-Based Spot GPU Instances;sustaining Performance While Reducing Energy Consumption: A Control Theory Approach;algorithm Design for Tensor Units;a Scalable Approximation Algorithm for Weighted Longest Common Subsequence;TSLQueue: An Efficient Lock-Free Design for Priority Queues;G-Morph: Induced Subgraph Isomorphism Search of Labeled Graphs on a GPU;accelerating Graph applications Using Phased Transactional Memory;Efficient GPU Computation Using Task Graph parallelism;towards High Performance Resilience Using Performance Portable Abstractions;Enhancing Load-Balancing of MPI applications with Workshare;trace-Based Workload Generation and Execution;particle-In-Cell Simulation Using Asynchronous Tasking;Exploiting Co-execution with OneAPI: Heterogeneity from a Modern Perspective;designing a 3D parallel Memory-Aware Lattice Boltzmann Algorithm on Manycore Systems;Fault-Tolerant LU Factorization Is Low Cost;Mixed Precision Incomplete and Factorized Sparse Approximate Inverse Preconditioning on GPUs;GPU-Accelerated Mahalanobis-Average Hierarchical Clustering Analysis.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Conceptual Framework for Next-Generation Software Ecosystems 22

Conceptual Framework for Next-Generation Software Ecosystems

引用

22nd IEEE/ACIS international conference on Software Engineering, Artificial Intelligence, Networking and parallel/distributed Computing (SNPD-Fall)

作者： Matsumoto, Kenichi Nara Inst Sci & Technol Div Informat Sci Ikoma Japan

ISBN: (纸本)9781665404037

This paper proposes a conceptual framework for developing new technologies that will solve today's technical issues in software development and operations (DevOps) and support the future software ecosystems. The proposed framework perceives resources essential for software DevOps from three perspectives: products, people, and technical information, and actively utilizes and link the latest digital technologies such as AI, natural language processing, microservices, and blockchain. The goal is not to aim fully automate software DevOps, but also to achieve high economic efficiency and sustainability by eliminating waste in software DevOps, assuming a human-centered society. The principal approaches of new technology development in the framework are "product up-cycling", "placement of the right people and AI in the right places", and "quality control linked to external technical information." New technologies to be developed with these approaches will expand conventional concepts in software DevOps with three dimensions of "reuse," "human resources," and "quality control."

关键词： Software Reuse Microservices AI applications Blockchain Technology

来源：评论

学校读者我要写书评

暂无评论

parallel Subtrajectory Alignment over Massive-Scale Trajectory Data 30

Parallel Subtrajectory Alignment over Massive-Scale Trajecto...

引用

30th international Joint conference on Artificial Intelligence (IJCAI)

作者： Chen, Lisi Shang, Shuo Feng, Shanshan Kalnis, Panos Univ Elect Sci & Technol China Chengdu Peoples R China Harbin Inst Technol Shenzhen Peoples R China King Abdullah Univ Sci & Technol Thuwal Saudi Arabia

ISBN: (纸本)9780999241196

We study the problem of subtrajectory alignment over massive-scale trajectory data. Given a collection of trajectories, a subtrajectory alignment query returns new targeted trajectories by splitting and aligning existing trajectories. The resulting functionality targets a range of applications, including trajectory data analysis, route planning and recommendation, ridesharing, and general location-based services. To enable efficient and effective subtrajectory alignment computation, we propose a novel search algorithm and filtering techniques that enable the use of the parallel processing capabilities of modern processors. Experiments with large trajectory datasets are conducted for evaluating the performance of our proposal. The results show that our solution to the subtrajectory alignment problem can generate high-quality results and are capable of achieving high efficiency and scalability.

关键词： Trajectories

来源：评论

学校读者我要写书评

暂无评论

GPGPU, multi-GPU, data movement, GPU cache management 31

GPGPU, multi-GPU, data movement, GPU cache management

引用

31st international conference on parallel Architectures and Compilation techniques, PACT 2022

作者： Tan, Xiaodan Serina Golikov, Pavel Vijaykumar, Nandita Pekhimenko, Gennady University of Toronto Vector Institute TorontoON Canada University of Toronto TorontoON Canada

ISBN: (纸本)9781450398688

As Graphics processing Units (GPUs) evolved into popular hardware accelerators for many compute-hungry applications in the cloud, GPU virtualization has become a highly desirable feature to lower operating costs of cloud infrastructures. In this work, we explore fne-grained GPU sharing, which has a larger potential in improving hardware utilization compared to a coarse-grained approach adopted in Nvidia's latest virtualization solution. However, fne-grained sharing introduces workload-dependent interference among concurrent applications, which leads to performance unpredictability and thus makes it extremely challenging to provide quality-of-service (QoS) guarantees in a cloud environment. To this end, we present GPUPool, which (i) unlocks the performance potential of fne-grained GPU sharing through mitigating interference and (ii) provides QoS guarantees to GPU workloads at both the systems and hardware levels. By leveraging our key insight that co-execution performance of GPU jobs is highly correlated with hardware utilization of individual workloads, GPUPool predicts performance of co-running incoming jobs and assigns compatible GPU jobs onto the same hardware. Given this performance estimation, GPUPool is able to satisfy job-specifc QoS requirements while maximizing hardware sharing across a GPU cluster. Our experimental results show GPUPool reduces the number of GPUs required to meet QoS requirements of incoming jobs by 31.2% and 21.1%, compared to a system with no sharing and one with coarse-grained sharing respectively. © 2022 Association for Computing Machinery.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Evolution of the computational science community: The dynamics of topics and collaborations in 24 years of ICCS and JoCS publications

引用

Journal of Computational Science 2025年 89卷

作者： Luo, Lijing Bochenina, Klavdiya Abuhay, Tesfamariam M. Dorzhu, Nachyn Kampis, George Kovalchuk, Sergey Krzhizhanovskaya, Valeria Paszynski, Maciej de Mulatier, Clélia Dongarra, Jack Sloot, Peter M.A. University of Amsterdam Amsterdam Netherlands University of Helsinki Helsinki Finland University of Illinois at Chicago ChicagoIL United States International Conference on Computational Science Amsterdam Netherlands Eötvös Loránd University Budapest Hungary AGH University of Krakow Krakow Poland University of Tennessee KnoxvilleTN United States

We analyze the topic structure of 10,299 publications from the international conference on Computational Science (ICCS) between 2001 and 2024 as well as the Journal of Computational Science (JoCS) between 2010 and 2023, using natural language processing techniques and network analysis. The computational science classification corpus was created into 15 main disciplines and 256 sub-disciplines sourced from Wikipedia. Among the 15 main disciplines, machine learning became the most popular topic after 2019, surpassing parallel & distributed computing, which peaked in the early 2010s. ICCS and JoCS show differences in research popularity in both first and second-level disciplines. Algorithm theory, Mathematical modeling, and network science are the most dominant topics in both ICCS and JoCS. Different disciplines present different trends in ICCS and JoCS. In the past 24 years, machine learning related topics have gained the most attention in both ICCS and JoCS. We also examined and compared the correlation between the trends in ICCS and Google search Trends. The collaboration of disciplinary networks of second-level disciplines exhibits a scale-free characteristic, and the network structures have undergone significant evolution over 24 years. Moreover, different disciplinary communities exhibit different "introverted" and "extroverted" community characteristics within the network. Additionally, we examined the life span of thematic workshops and the evolution of authors’ collaborations inside and after ICCS. © 2025 The Authors

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Dynamic Batch parallel Algorithms for Updating PageRank

Dynamic Batch Parallel Algorithms for Updating PageRank

引用

IEEE international Symposium on parallel and distributed processing Workshops and Phd Forum (IPDPSW)

作者： Subhajit Sahu Kishore Kothapalli Dip Sankar Banerjee International Institute of Information Technology Hyderabad India Indian Institute of Technology Jodhpur India

ISBN: (数字)9781665497473

ISBN: (纸本)9781665497480

The design and implementation of parallel algorithms for dynamic graph problems is attracting significant research attention in the recent years, driven by numerous applications to social network analysis, neuroscience, and protein interaction networks. One such problem is the computation of PageRank values of vertices in a directed graph. This paper presents two new parallel algorithms for recomputing the PageRank values of vertices in a dynamic graph. Our techniques require the recomputation of the PageRank of only the vertices affected by the insertion/deletion of a batch of edges. We conduct detailed experimental studies of our algorithm on a set of 11 real-world graphs. Our results on Intel Xeon Silver 4116 CPU and NVIDIA Tesla V100 PCIe 16GB GPU indicate that our algorithms outperform static and dynamic update algorithms by $6.1\times$ : and $8.6\times \mathbf{on}$ the CPU, and by 9.8×and $9.3\times\mathbf{on}$ the GPU respectively. We also compare the performance of the algorithms in batched mode to cumulative single-edge updates.

关键词： Proteins Silver distributed processing Neuroscience Social networking (online) Heuristic algorithms Graphics processing units

来源：评论

学校读者我要写书评

暂无评论

QR Factorization of Block Low-Rank Matrices on Multi-instance GPU 23rd

QR Factorization of Block Low-Rank Matrices on Multi-insta...

引用

23rd international conference on parallel and distributed Computing, applications, and Technologies, PDCAT 2022

作者： Ohshima, Satoshi Ida, Akihiro Yokota, Rio Yamazaki, Ichitaro Information Technology Center Nagoya University Aichi Japan Research Institute for Value-Added-Information Generation Japan Agency for Marine-Earth Science and Technology Kanagawa Japan Global Scientific Information and Computing Center Tokyo Institute of Technology Tokyo Japan Scalable Algorithms Department Sandia National Laboratories New Mexico United States

ISBN: (纸本)9783031299261

The QR factorization, which is a fundamental operation in linear algebra, is used extensively in scientific simulations. The acceleration and memory reduction of it are important research targets. QR factorization using block low-rank matrices (BLR-QR) has previously been proposed to address this issue. In this study, we consider its implementation on a GPU. Current CPUs and GPUs have numerous computational cores and the performance consists of the total performance of them. Therefore, the degree of parallelism of the target calculation is important for obtaining high performance. By contrast, many applications, including BLR-QR, do not have sufficient parallelism. Batched computation has attracted attention for achieving high performance in such calculations. However, the use of it requires major code rewriting and is extremely laborious. Thus, we propose the use of the multi-instance GPU (MIG) feature of current GPUs. Using MIG, we succeeded in obtaining a 53.3% time reduction over the CPU and 77.6% over the GPU without MIG. From the above result, we succeeded in demonstrating rapid implementation of BLR-QR on MIG and usefulness of MIG. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

SALoBa: Maximizing Data Locality and Workload Balance for Fast Sequence Alignment on GPUs

SALoBa: Maximizing Data Locality and Workload Balance for Fa...

引用

international Symposium on parallel and distributed processing (IPDPS)

作者： Seongyeon Park Hajin Kim Tanveer Ahmad Nauman Ahmed Zaid Al-Ars H. Peter Hofstee Youngsok Kim Jinho Lee Yonsei University TU Delft IBM

Sequence alignment forms an important backbone in many sequencing applications. A commonly used strategy for sequence alignment is an approximate string matching with a two-dimensional dynamic programming approach. Although some prior work has been conducted on GPU acceleration of a sequence alignment, we identify several shortcomings that limit exploiting the full computational capability of modern GPUs. This paper presents SALoBa, a GPU-accelerated sequence alignment library focused on seed extension. Based on the analysis of previous work with real-world sequencing data, we propose techniques to exploit the data locality and improve work-load balancing. The experimental results reveal that SALoBa significantly improves the seed extension kernel compared to state-of-the-art GPU-based methods.

关键词： Performance evaluation Sequential analysis distributed processing Graphics processing units Genomics Performance gain Software

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：