检索结果-内蒙古大学图书馆

IEEE COMPUTER ARCHITECTURE LETTERS 2024年第1期23卷 125-128页

作者： Qi, Yingjie Yang, Jianlei Zhou, Ao Qiao, Tong Hu, Chunming Beihang Univ Sch Comp Sci & Engn Beijing 100191 Peoples R China

Graph neural networks (GNNs) have gained significant popularity due to the powerful capability to extract useful representations from graph data. As the need for efficient GNN computation intensifies, a variety of programming abstractions designed for optimizing GNN Aggregation have emerged to facilitate acceleration. However, there is no comprehensive evaluation and analysis upon existing abstractions, thus no clear consensus on which approach is better. In this letter, we classify existing programming abstractions for GNN Aggregation by the dimension of data organization and propagation method. By constructing these abstractions on a state-of-the-art GNN library, we perform a thorough and detailed characterization study to compare their performance and efficiency, and provide several insights on future GNN acceleration based on our analysis.

关键词： programming Graph neural networks Organizations Kernel Indexes Graphics processing units Taxonomy Graph neural networks (GNNs) characterization programming abstractions execution patterns

来源：评论

学校读者我要写书评

暂无评论

programming abstractions for Software-Defined Wireless Networks

引用

IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT 2015年第2期12卷 146-162页

作者： Riggio, Roberto Marina, Mahesh K. Schulz-Zander, Julius Kuklinski, Slawomir Rasheed, Tinku CREATE NET I-38123 Trento Italy Univ Edinburgh Sch Informat Edinburgh EH8 9Y Midlothian Scotland Tech Univ Berlin TU Berlin D-10623 Berlin Germany Orange Polska PL-96100 Warsaw Poland

Software-Defined Networking (SDN) has received, in the last years, significant interest from the academic and the industrial communities alike. The decoupled control and data planes found in an SDN allows for logically centralized intelligence in the control plane and generalized network hardware in the data plane. Although the current SDN ecosystem provides a rich support for wired packet-switched networks, the same cannot be said for wireless networks where specific radio data-plane abstractions, controllers, and programming primitives are still yet to be established. In this work, we present a set of programming abstractions modeling the fundamental aspects of a wireless network, namely state management, resource provisioning, network monitoring, and network reconfiguration. The proposed abstractions hide away the implementation details of the underlying wireless technology providing programmers with expressive tools to control the state of the network. We also present a Software-Defined Radio Access Network Controller for Enterprise WLANs and a Python-based Software Development Kit implementing the proposed abstractions. Finally, we experimentally evaluate the usefulness, efficiency and flexibility of the platform over a real 802.11-based WLAN.

关键词： Network management programming abstractions software-defined wireless networks WLANs

来源：评论

学校读者我要写书评

暂无评论

SPGPU: Spatially Programmed GPU

IEEE COMPUTER ARCHITECTURE LETTERS

引用

IEEE COMPUTER ARCHITECTURE LETTERS 2024年第2期23卷 223-226页

作者： Zhu, Shizhuo Shkirko, Illia Levinson, Jacob Wang, Zhengrong Nowatzki, Tony Univ Calif Los Angeles Los Angeles CA 90095 USA

Communication is a critical bottleneck for GPUs, manifesting as energy and performance overheads due to network-on-chip (NoC) delay and congestion. While many algorithms exhibit locality among thread blocks and accessed data, modern GPUs lack the interface to exploit this locality: GPU thread blocks are mapped to cores obliviously. In this work, we explore a simple extension to the conventional GPU programming interface to enable control over the spatial placement of data and threads, yielding new opportunities for aggressive locality optimizations within a GPU kernel. Across 7 workloads that can take advantage of these optimizations, for a 32 (or 128) SM GPU: we achieve a 1.28x (1.54x) speedup and 35% (44%) reduction in NoC traffic, compared to baseline non-spatial GPUs.

关键词： Instruction sets Graphics processing units Message systems programming Topology Hardware Kernel Optimization Registers Pipelines Spatial architectures GPU programming abstractions shared-data locality on-chip networks

来源：评论

学校读者我要写书评

暂无评论

Towards Real-Time Aggregate Computing 12th

Towards Real-Time Aggregate Computing

引用

12th International Symposium on Leveraging Applications of Formal Methods, Verification and Validation

作者： Audrito, Giorgio Damiani, Ferruccio Torta, Gianluca Univ Torino Turin Italy

ISBN: (纸本)9783031751066;9783031751073

Developing large-scale collective adaptive systems for safety-critical applications requires an extensive effort, involving the interplay of distributed programming techniques and mathematical proofs of real-time guarantees. This effort could be significantly reduced by allowing the system developer to rely on libraries of predefined algorithms. By exploiting such algorithms, distributed behaviour and (hard) real-time guarantees for the final application could be automatically inferred, effectively shifting the verification burden from the system designer to the algorithm developer. Following earlier work on real-time guarantees for aggregate computing algorithms, we argue that aggregate computing could provide a convenient framework towards this aim. As a first step, we give a detailed description of different kinds of models that abstract aggregate programs as mathematical functions. Then, building on such models, we start investigating the problem of how real-time behavior constraints could be specified in a compositional way. Finally, we conclude by singling out a number of potential building block library algorithms that could constitute such a real-time aggregate computing library, with the potential of providing a valuable asset for supporting the rigorous engineering of safety-critical large-scale collective adaptive systems.

关键词： Self-organising systems Real-time guarantees Aggregate Computing programming abstractions

来源：评论

学校读者我要写书评

暂无评论

MPR: An MPI Framework for Distributed Self-adaptive Stream Processing 30th

MPR: An MPI Framework for Distributed Self-adaptive Stream P...

引用

30th European Conference on Parallel and Distributed Processing (Euro-Par)

作者： Loff, Junior Griebler, Dalvan Fernandes, Luiz Gustavo Binder, Walter Univ Svizzera Italiana USI Fac Informat Lugano Switzerland Pontif Catholic Univ Rio Grande do Sul PUCRS Sch Technol Porto Alegre RS Brazil

ISBN: (纸本)9783031695827;9783031695834

Stream processing systems must often cope with workloads varying in content, format, size, and input rate. The high variability and unpredictability make statically fine-tuning them very challenging. Our work addresses this limitation by providing a new framework and run-time system to simplify implementing and assessing new self-adaptive algorithms and optimizations. We implement a prototype on top of MPI called MPR and show its functionality. We focus on horizontal scaling by supporting the addition and removal of processes during execution time. Experiments reveal that MPR can achieve performance similar to that of a handwritten static MPI application. We also assess MPR's adaptation capabilities, showing that it can readily re-configure itself, with the help of a self-adaptive algorithm, in response to workload variations.

关键词： Stream Processing Distributed Systems Self-adaptive Parallel programming programming abstractions MPI

来源：评论

学校读者我要写书评

暂无评论

Towards Higher-Level abstractions for Quantum Computing 22

Towards Higher-Level Abstractions for Quantum Computing

引用

Australian Computer Science Week (ACSW)

作者： Cobb, Adrian Schneider, Jean-Guy Lee, Kevin Deakin Univ Geelong Vic Australia

ISBN: (纸本)9781450396066

Quantum Computing (QC) has emerged as a field of ever-increasing activity as it promises to revolutionize computation and enable the solution of computational problems that we (realistically) cannot solve with Classical Computing to date. However, existing quantum programming environments mostly require an in-depth understanding of the basic QC building blocks, that is, quantum states, superposition, entanglement and measurement as well as the changing of quantum states using basic quantum gates. The present state of quantum programming reminds us of how Classical Computing was about 70 years ago when computing machines such as the ENIAC required significant effort to program solely using very basic digital building blocks. Over the decades, though, increasingly higher-level abstractions have been created on top of the basic building blocks of Classical Computing and made computation much more accessible and wide-spread. In order to make Quantum Computing more accessible, we argue that Software Engineering for QC needs to embark on a similar journey and create abstractions that shield developers from the basic QC building blocks as much as possible so that they can focus their attention on solving problems and less on how to manipulate quantum states using quantum circuits. Based on our experience of developing a scalable quantum n-queens solver, this paper aims to formulate recommendations for raising the level of abstraction in Quantum Software Engineering.

关键词： Quantum Computing Quantum Software Engineering programming abstractions

来源：评论

学校读者我要写书评

暂无评论

EXPLOITING WIRELESS LINKS DIVERSITY IN SOFTWARE-DEFINED IEEE 802.11 ENTERPRISE WLAN

EXPLOITING WIRELESS LINKS DIVERSITY IN SOFTWARE-DEFINED IEEE...

引用

IEEE Global Communications Conference (GLOBECOM) on Advanced Technology for 5G Plus

作者： Monir, Md Fahad Granelli, Fabrizio Independent Univ Bangladesh Dept Comp Sci & Engn Dhaka Bangladesh Univ Trento UniTrento Dept Informat Engn & Comp Sci Trento Italy

ISBN: (纸本)9781728182988

Adoption of wireless technologies is rising dramatically. As traffic load is continuously rising with the rapid growth of user scale and mobile services, today's Wireless Local Area Network (WLAN) enterprise is facing a series of crucial challenges such as packet loss, interference, poor bandwidth etc. This is because of the inherent design of the IEEE 802.11 architecture. Enabling Software Defined Networking (SDN) in Wi-Fi enterprise would solve the puzzle of current WLAN technology limitations and ensure high scalability and performance. While standardization is focusing on performance enhancement by evolving the Wi-F1 MAC/PHY protocols, in this work we focus on bandwidth optimization in IEEE 802.11 by using Software Defined Networking (SDN). We analyze various programming abstractions for WLANs and exploit the wireless link diversity in Software Defined Enterprise-WLAN with the support of the 5GEMPOWERSDN platform. A multiple uplink mechanism between user devices and Access Points (APs) is proposed that allows a remarkable performance improvement in the network in terms of throughput and packet delivery ratio.

关键词： Software Defined Networking (SDN) Software Defined Wireless Networks (SDWN) WLANs network performance programming abstractions

来源：评论

学校读者我要写书评

暂无评论

Trends in Data Locality abstractions for HPC Systems

引用

IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS 2017年第10期28卷 3007-3020页

作者： Unat, Didem Dubey, Anshu Hoefler, Torsten Shalf, John Abraham, Mark Bianco, Mauro Chamberlain, Bradford L. Cledat, Romain Edwards, H. Carter Finkel, Hal Fuerlinger, Karl Hannig, Frank Jeannot, Emmanuel Kamil, Amir Keasler, Jeff Kelly, Paul H. J. Leung, Vitus Ltaief, Hatem Maruyama, Naoya Newburn, Chris J. Pericas, Miquel Koc Univ Dept Comp Engn TR-34450 Istanbul Turkey Argonne Natl Lab 9700 S Cass Ave Argonne IL 60439 USA ETH CH-8092 Zurich Switzerland Lawrence Berkeley Natl Lab Berkeley CA 94720 USA KTH Royal Inst Technol S-17121 Solna Sweden Swiss Natl Supercomp Ctr CH-6900 Lugano Switzerland Cray Inc Seattle WA 98164 USA Intel Cooperat Santa Clara CA 95050 USA Sandia Natl Labs POB 5800 Albuquerque NM 87185 USA Argonne Natl Lab 9700 S Cass Ave Argonne IL 60439 USA Ludwig Maximilians Univ Munchen D-80538 Munich Germany Univ Erlangen Nurnberg D-91058 Erlangen Germany INRIA Bordeaux Sud Ouest F-33405 Talence France Univ Michigan Ann Arbor MI 48109 USA Lawrence Berkeley Natl Lab Berkeley CA 94720 USA Lawrence Livermore Natl Lab Livermore CA 94550 USA Imperial Coll London Software Technol London England King Abdullah Univ Sci & Technol Thuwal 23955 Saudi Arabia RIKEN Kobe Hyogo 6500047 Japan Nvidia Corp Santa Clara CA 95050 USA Chalmers Univ Technol S-41296 Gothenburg Sweden

The cost of data movement has always been an important concern in high performance computing (HPC) systems. It has now become the dominant factor in terms of both energy consumption and performance. Support for expression of data locality has been explored in the past, but those efforts have had only modest success in being adopted in HPC applications for various reasons. them However, with the increasing complexity of the memory hierarchy and higher parallelism in emerging HPC systems, locality management has acquired a new urgency. Developers can no longer limit themselves to low-level solutions and ignore the potential for productivity and performance portability obtained by using locality abstractions. Fortunately, the trend emerging in recent literature on the topic alleviates many of the concerns that got in the way of their adoption by application developers. Data locality abstractions are available in the forms of libraries, data structures, languages and runtime systems;a common theme is increasing productivity without sacrificing performance. This paper examines these trends and identifies commonalities that can combine various locality concepts to develop a comprehensive approach to expressing and managing data locality on future large-scale high-performance computing systems.

关键词： Data locality programming abstractions high-performance computing data layout locality-aware runtimes

来源：评论

学校读者我要写书评

暂无评论

Signal/Collect Processing large graphs in seconds

引用

SEMANTIC WEB 2016年第2期7卷 139-166页

作者： Stutz, Philip Strebel, Daniel Bernstein, Abraham Univ Zurich CH-8006 Zurich Switzerland

Both researchers and industry are confronted with the need to process increasingly large amounts of data, much of which has a natural graph representation. Some use MapReduce for scalable processing, but this abstraction is not designed for graphs and has shortcomings when it comes to both iterative and asynchronous processing, which are particularly important for graph algorithms. This paper presents the Signal/Collect programming model for scalable synchronous and asynchronous graph processing. We show that this abstraction can capture the essence of many algorithms on graphs in a concise and elegant way by giving Signal/Collect adaptations of algorithms that solve tasks as varied as clustering, inferencing, ranking, classification, constraint optimisation, and even query processing. Furthermore, we built and evaluated a parallel and distributed framework that executes algorithms in our programming model. We empirically show that our framework efficiently and scalably parallelises and distributes algorithms that are expressed in the programming model. We also show that asynchronicity can speed up execution times. Our framework can compute a PageRank on a large (>1.4 billion vertices, >6.6 billion edges) real-world graph in 112 seconds on eight machines, which is competitive with other graph processing approaches.

关键词： Distributed computing scalability programming abstractions programming models graph processing

来源：评论

学校读者我要写书评

暂无评论

Dynamic Sets - A programming Abstraction for Ubiquitous Computing and the Internet of Things 1

Dynamic Sets - A Programming Abstraction for Ubiquitous Comp...

引用

1st International Workshop on Mashups of Things and APIs (MoTA)

作者： Prellwitz, Matthias Parzyjegla, Helge Muehl, Gero Timmermann, Dirk Univ Rostock Rostock Germany

ISBN: (纸本)9781450346696

Applications in the areas of Ubiquitous Computing and the Internet of Things are often facing the problem that at design time it is not known what devices will be available at runtime and what preferences the user will have. Due to this gap between design time and runtime, mechanisms are needed to postpone decisions until the application is deployed and to adapt these decisions while it is running. Unfortunately, such mechanisms are currently mostly missing. In this paper, we propose dynamic sets as a programming abstraction that allows an application originally written for using a single device to interact with a set of devices of the same type instead. Using a dynamic set, an application can transparently call a method on a set of devices using a local proxy of the set that implements the very same interface as a device the application was originally written to interact with. Applications that are aware of dynamic sets can change the criteria used to select the members of a set or derive a new set from existing sets. Besides invocation mechanisms, dynamic sets provide functionality for result aggregation, automatic member management, update notifications, and derived business methods. Finally, applications can be exported making it possible to combine existing applications to flexible and dynamic mashups.

关键词： programming abstractions object grouping

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：