检索结果-南通市图书馆

D-Cubicle:boosting data transfer dynamically for large-scale analytical queries in single-GPU systems

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer science 2023年第4期17卷 141-153页

作者： Jialun WANG Wenhao PANG Chuliang WENG Aoying ZHOU School of Data Science and Engineering East China Normal UniversityShanghai 200062China

In analytical queries,a number of important operators like JOIN and GROUP BY are suitable for parallelization,and GPU is an ideal accelerator considering its power of parallel computing.However,when data size increases to hundreds of gigabytes,one GPU card becomes insufficient due to the small capacity of global memory and the slow data transfer between host and device.A straightforward solution is to equip more GPUs linked with high-bandwidth connectors,but the cost will be highly increased.We utilize unified memory(UM)produced by NVIDIA CUDA(Compute Unified Device Architecture)to make it possible to accelerate large-scale queries on just one GPU,but we notice that the transfer performance between host and UM,which happens before kernel execution,is often significantly slower than the theoretical bandwidth.An important reason is that,in singleGPU environment,data processing systems usually invoke only one or a static number of threads for data copy,leading to an inefficient transfer which slows down the overall performance heavily.In this paper,we present D-Cubicle,a runtime module to accelerate data transfer between host-managed memory and unified memory.D-Cubicle boosts the actual transfer speed dynamically through a self-adaptive approach.In our experiments,taking data transfer into account,D-Cubicle processes 200 GB of data on a single GPU with 32 GB of global memory,achieving 1.43x averagely and 2.09x maximally the performance of the baseline system.

关键词： data analytics GPU unified memory

Hetu: a highly efficient automatic parallel distributed deep learning system

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

science China(Information sciences) 2023年第1期66卷 314-315页

作者： Xupeng MIAO Xiaonan NIE Hailin ZHANG Tong ZHAO Bin CUI school of Computer science Peking University Center for data science Peking University National engineering Laboratory for Big data Analysis and Applications

On July 18, 2021, the PKU-DAIR Lab1)(data and Intelligence Research Lab at Peking University) openly released the source code of Hetu, a highly efficient and easy-to-use distributed deep learning(DL) framework. Hetu is the first distributed DL system developed by academic groups in Chinese universities, and takes into account both high availability in industry and innovation in academia. Through independent research and development, Hetu is completely decoupled from the existing DL systems and has unique characteristics. The public release of the Hetu system will help researchers and practitioners to carry out frontier MLSys(machine learning system) research and promote innovation and industrial upgrading.

关键词：

Dynamic depth-width optimization for capsule graph convolutional network

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer science 2023年第6期17卷 159-161页

作者： Shangwei WU Yingtong XIONG Chuliang WENG School of Data Science and Engineering East China Normal UniversityShanghai 200062China

1 Introduction Encouraged by the success of Convolutional Neural Networks(CNNs),many studies[1],known as Graph Convolutional Networks(GCNs),borrowed the idea of convolution and redefined it for graph data.In graph-level classification tasks,Classic GCN methods[2,3]generate graph embeddings based on the learned node embeddings which consider each node’s representation as multiple independent scalar features.However,they neglect the detailed mutual relations among different node features such as position,direction,and connection.Inspired by CapsNet[4]which encodes each feature of an image as a vector(a capsule),CapsGNN[5]extracts multi-scale node features from different convolutional layers in the form of capsules.However,CapsGNN uses a static model structure to conduct training,which inherently restricts its representation ability on different datasets.

关键词： convolution representation mutual

BATON: symphony of random testing and concolic testing through machine learning and taint analysis

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

science China(Information sciences) 2023年第3期66卷 97-117页

作者： Bihuan CHEN Yang LIU Xin PENG Yijian WU Shengchao QIN school of Computer science Fudan University Shanghai Key Laboratory of data science Fudan University school of Computer science and engineering Nanyang Technological University school of Computing Teesside University

Random testing is scalable but often fails to hit corner program behaviors,while systematic testing (e.g.,concolic execution) is promising to cover corner program behaviors but is not scalable to explore all program behaviors.Prior attempts to integrate random testing with systematic testing lack targeted guidance.In this paper,we propose a guided hybrid testing approach,named BATON,to synergize random testing with concolic testing.It integrates the knowledge inside test cases and their executions into a conditional execution graph,and uses such knowledge to guide test case generation.Specifically,we learn classification models for some conditionals in the conditional execution graph in a demand-driven way.These models are used to guide random testing to reach and cover partially-covered conditionals.We further employ targeted concolic testing to cover conditionals that cannot be fully covered by guided random testing.We implemented BATONfor Java and evaluated it on three benchmarks.The results show that BATONimproved branch coverage and mutation score over random testing by 16.2%–29.4%and 19.0%–30.0%,over adaptive random testing by 16.8%–33.8%and 19.4%–34.2%,over concolic testing by 2.3%–29.9%and 2.9%–30.1%,and over simple hybrid testing by 1.6%–14.5%and 1.4%–18.7%.

关键词： system testing random testing concolic testing

Key-Value Store Coupled with an Operating System for Storing Large-Scale Values

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2022年第11期73卷 3333-3350页

作者： Jeonghwan Im Hyuk-Yoon Kwon Graduate school of data science Seoul National University of Science and TechnologySeoulKorea Department of Industrial engineering Graduate School of Data ScienceResearch Center for Electrical and Information ScienceSeoul National University of Science and TechnologySeoulKorea

The key-value store can provide flexibility of data types because it does not need to specify the data types to be stored in advance and can store any types of data as the value of the key-value pair.Various types of studies have been conducted to improve the performance of the key-value store while maintaining its flexibility.However,the research efforts storing the large-scale values such as multimedia data files(e.g.,images or videos)in the key-value store were limited.In this study,we propose a new key-value store,WR-Store++aiming to store the large-scale values stably.Specifically,it provides a new design of separating data and index by working with the built-in data structure of the Windows operating system and the file system.The utilization of the built-in data structure of the Windows operating system achieves the efficiency of the key-value store and that of the file system extends the limited space of the storage significantly.We also present chunk-based memory management and parallel processing of WR-Store++to further improve its performance in the GET operation.Through the experiments,we show that WR-Store++can store at least 32.74 times larger datasets than the existing baseline key-value store,WR-Store,which has the limitation in storing large-scale data sets.Furthermore,in terms of processing efficiency,we show that WR-Store++outperforms not only WR-Store but also the other state-ofthe-art key-value stores,LevelDB,RocksDB,and BerkeleyDB,for individual key-value operations and mixed workloads.

关键词： Key-value stores large-scale values chunk-based memory management parallel processing

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer science 2020年第1期14卷 191-207页

作者： Huiping LIU Cheqing JIN Aoying ZHOU School of Data Science and Engineering School of Computer Science and Software EngineeringEast China Normal UniversityShanghai 200062China

With the increasing number of GPS-equipped vehicles,more and more trajectories are generated continuously,based on which some urban applications become feasible,such as route planning.In general,popular route that has been travelled frequently is a good choice,especially for people who are not familiar with the road networks.Moreover,accurate estimation of the travel cost(such as travel time,travel fee and fuel consumption)will benefit a wellscheduled trip plan.In this paper,we address this issue by finding the popular route with travel cost estimation.To this end,we design a system consists of three main components.First,we propose a novel structure,called popular traverse graph where each node is a popular location and each edge is a popular route between locations,to summarize historical trajectories without road network information.Second,we propose a self-adaptive method to model the travel cost on each popular route at different time interval,so that each time interval has a stable travel cost.Finally,based on the graph,given a query consists of source,destination and leaving time,we devise an efficient route planning algorithmwhich considers optimal route concatenation to search the popular route from source to destination at the leaving time with accurate travel cost estimation.Moreover,we conduct comprehensive experiments and implement our system by a mobile App,the results show that our method is both effective and efficient.

关键词： location-based services route planning travel cost estimation minimum description length optimal road concatenation

Internet of Brain, Thought, Thinking, and Creation

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2022年第6期31卷 1025-1042页

作者： ZHANG Zhimin YIN Rui NING Huansheng school of Computer and Communication engineering University of Science and Technology Beijing Beijing engineering Research Center for Cyberspace data Analysis and Applications

Thinking space came into being with the emergence of human civilization. With the emergence and development of cyberspace, the interaction between those two spaces began to take place. In the collision of thinking and technology, new changes have taken place in both thinking space and cyberspace. To this end, this paper divides the current integration and development of thinking space and cyberspace into three stages, namely Internet of brain(IoB), Internet of thought(IoTh), and Internet of thinking(IoTk). At each stage, the contents and technologies to achieve convergence and connection of spaces are discussed. Besides, the Internet of creation(IoC) is proposed to represent the future development of thinking space and cyberspace. Finally, a series of open issues are raised, and they will become thorny factors in the development of the Io C stage.

关键词： Thinking space Cyberspace Internet of brain (IoB) Internet of thought (IoTh) Internet of thinking (IoTk) Internet of creation (IoC)

Probe Machine Based Computing Model for Maximum Clique Problem

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2022年第2期31卷 304-312页

作者： CUI Jianzhong YIN Zhixiang TANG Zhen YANG Jing Department of Computer Huainan Union University school of Electronic and Information engineering Anhui University of Science & Technology school of Mathematics Physics and Statistics Shanghai University of Engineering Science school of Mathematics and Big data Anhui University of Science & Technology

Probe machine(PM) is a recently reported mathematic model with massive parallelism. Herein,we presented searching the maximum clique of an undirected graph with six vertices. We constructed data library containing n sublibraries, each sublibrary corresponded to a vertex in the given graph. Then, probe library according to the induced subgraph was designed in order to search and generate all maximal cliques. Subsequently,we performed probe operation, and all maximal cliques were generated in parallel. The advantages of the proposed model lie in two aspects. On one hand, solution to NP-complete problem is generated in just one step of probe operation rather than found in vast solution space.On the other hand, the proposed model is highly parallel.The work demonstrates that PM is superior to TM in terms of searching capacity when tackling NP-complete problem.

关键词： Probe machine NP-complete Maximum clique problem

Resource investment for DDoS attack resistant SDN: a practical assessment

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science China(Information sciences) 2023年第7期66卷 112-129页

作者： Bin YUAN Fan ZHANG Jun WAN Huan ZHAO Shui YU Deqing ZOU Qiangsheng HUA Hai JIN school of Cyber science and engineering Huazhong University of Science and Technology National engineering Research Center for Big data Technology and System Services Computing Technology and System Lab Hubei engineering Research Center on Big data Security Shenzhen Huazhong University of science and Technology Research Institute school of Computer science and Technology Huazhong University of Science and Technology Cluster and Grid Computing Lab school of Computer science University of Technology Sydney

Software-defined networks(SDNs) present a novel network architecture that is widely used in various datacenters. However, SDNs also suffer from many types of security threats, among which a distributed denial of service(DDoS) attack, which aims to drain the resources of SDN switches and controllers,is one of the most common. Once the switch or controller is damaged, the network services can be affected.Many defense schemes against DDoS attacks have been proposed from the perspective of attack detection;however, such defense schemes are known to suffer from a time consuming and unpromising accuracy, which could result in an unavailable network service before specific countermeasures are taken. To address this issue through a systematic investigation, we propose an elaborate resource-management mechanism against DDoS attacks in an SDN. Specifically, by considering the SDN topology, we leverage the M/M/c queuing model to measure the resistance of an SDN to DDoS attacks. Network administrators can therefore invest a reasonable number of resources into SDN switches and SDN controllers to defend against DDoS attacks while guaranteeing the quality of service(QoS). Comprehensive analyses and empirical data-based experiments demonstrate the effectiveness of the proposed approach.

关键词： SDN security DDoS attacks resource management queueing theory QoS