2020 Eighth International Symposium on Computing and Networking (CANDAR)最新文献

Performance Improvement of Hadoop ext4-based Disk I/O 基于Hadoop ext4的磁盘I/O性能提升

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00032

Makoto Nakagami, J. Fortes, Saneyasu Yamaguchi

{"title":"Performance Improvement of Hadoop ext4-based Disk I/O","authors":"Makoto Nakagami, J. Fortes, Saneyasu Yamaguchi","doi":"10.1109/CANDAR51075.2020.00032","DOIUrl":"https://doi.org/10.1109/CANDAR51075.2020.00032","url":null,"abstract":"Hadoop is one of the most popular big-data analytics platforms, often relying on hard disk drives for storage of big-data amounts that exceed the capacity of solid-state drives. Unlike other data-intensive applications, such as database management systems, big-data processing jobs frequently require extensive sequential I/O requests. Previously proposed methods for improving sequential I/O performance modified the block usage bitmap of the Ext2/3 filesystem in order to actively use the faster disk zones, which are the outer zones in each hard disk drive. However, these methods do not support Ext4, which is the current version of Ext filesystems. In this paper, we discuss a method for improving the sequential I/O performance of the Ext4 filesystem. First, we evaluate the sequential file access throughputs on Ext3, Ext4, and XFS filesystems. We point out that Ext4 does not actively utilize the area freed by deleting existing files, causing declines in file access performance. Second, we propose a method for improving the Ext4 sequential file access performance. The improved Ext4 actively utilizes the faster zones of storage devices by controlling file placement location. Third, we evaluate the proposed filesystem and show that it outperforms existing filesystems. In the case of TeraSort, Hadoop with the proposed Ext4 filesystem performs better than when using the original Ext4 filesystem by as much as 30.1%.","PeriodicalId":215489,"journal":{"name":"2020 Eighth International Symposium on Computing and Networking (CANDAR)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116776417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Virtual Machine Monitor-based Hiding Method for Access to Debug Registers 基于虚拟机监视器的隐藏调试寄存器访问方法

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00036

Masaya Sato, H. Taniguchi, Ryosuke Nakamura

引用次数: 2

On the Loop Suppression Method by Utilizing Information of Master Terminals for Neighboring Terminals in Bluetooth MANETs 蓝牙无线局域网中利用主终端信息抑制相邻终端环路的方法研究

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00025

Haruki Ishizaki, Ryohei Saka, Eitaro Kohno, Y. Kakuda

{"title":"On the Loop Suppression Method by Utilizing Information of Master Terminals for Neighboring Terminals in Bluetooth MANETs","authors":"Haruki Ishizaki, Ryohei Saka, Eitaro Kohno, Y. Kakuda","doi":"10.1109/CANDAR51075.2020.00025","DOIUrl":"https://doi.org/10.1109/CANDAR51075.2020.00025","url":null,"abstract":"Bluetooth MANETs are ad hoc networks consist of mobile terminals using Bluetooth. Recently, a fast connection-establishment method between terminals in piconets has been proposed. In the existing method, however, as the density of terminals increases, the existing method can generate many connections between the slaves in the piconets and can easily form loops. Since Bluetooth MANETs utilize flooding for data packet forwarding, the amount of data packets increases due to the loops. In this paper, we propose a method to control excessive connection-establishment between slave terminals in piconets for Bluetooth MANETs to suppress the loops. In order to suppress the loops, our proposed method considers information of master terminals for neighboring terminals. From the simulation experiments, we have confirmed that our proposed method can suppress the number of loops regardless of the terminal density, and our proposed method also can suppress the amount of data packets while keeping the time to complete the dissemination of data packets at the same level as the existing method.","PeriodicalId":215489,"journal":{"name":"2020 Eighth International Symposium on Computing and Networking (CANDAR)","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134545637","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ExtraFerns: Fully Parallel Ensemble Learning Technique with Non-Greedy yet Minimal Memory Access Training 外置:完全并行集成学习技术与非贪婪，但最小的内存访问训练

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00027

Shungo Kumazawa, Kazushi Kawamura, Thiem Van Chu, M. Motomura, Jaehoon Yu

引用次数: 1

Bandit-based Run-time Adaptation of Cache Replacement Policies in Content Management Systems 内容管理系统中基于强盗的缓存替换策略的运行时自适应

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00017

Srivattsan Sridharan, Shravan Gaonkar, J. Fortes

{"title":"Bandit-based Run-time Adaptation of Cache Replacement Policies in Content Management Systems","authors":"Srivattsan Sridharan, Shravan Gaonkar, J. Fortes","doi":"10.1109/CANDAR51075.2020.00017","DOIUrl":"https://doi.org/10.1109/CANDAR51075.2020.00017","url":null,"abstract":"Web traffic is increasingly complex, in terms of both content accessed and geographical distribution of requests. Content providers usually employ geographically distributed web-caches to reduce latency in delivering content. A web-cache configured with a single replacement policy cannot provide optimal performance due to the inherent variability associated with the request pattern. We describe an adaptation mechanism (DETACH) that utilizes a multi-armed bandit approach to adapt the replacement policy depending on a new metric, the rank of evicted objects, which indirectly characterizes run-time request patterns. The proposed mechanism requires very minimal modifications to existing cache systems and does not maintain any history (ghost entries). We also describe the implementation of a cache system that uses the proposed mechanism to adapt the replacement policy at runtime. We consider two replacement policies, LRU and LFU, but the system is expandable to other policies. Our experiments with various datasets show that the system adapts quickly to the best performing replacement policy. We compare the performance of the system to a recently proposed approach called LeCaR, and baseline LRU and LFU policies. Of the 7 datasets used on our experiments, DETACH outperformed LeCaR in 4 of them by up to a factor of two. In the remaining 3 datasets LeCaR outperformed DETACH by up to 5%. Our proposed system consistently outperforms the baseline LRU and LFU replacement policies for all the datasets.","PeriodicalId":215489,"journal":{"name":"2020 Eighth International Symposium on Computing and Networking (CANDAR)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127136627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Use cases and success stories of a data analytics system in an automotive Paint Shop 汽车喷漆车间数据分析系统的用例和成功案例

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00019

Ángel Dacal-Nieto, J. J. Areal, Miguel García-Fernández, Marcos Lluch

{"title":"Use cases and success stories of a data analytics system in an automotive Paint Shop","authors":"Ángel Dacal-Nieto, J. J. Areal, Miguel García-Fernández, Marcos Lluch","doi":"10.1109/CANDAR51075.2020.00019","DOIUrl":"https://doi.org/10.1109/CANDAR51075.2020.00019","url":null,"abstract":"Manufacturing companies are currently developing transformation plans from the 3.0 paradigm, based on automation, robotics and Lean, to the 4.0 approach, based on digitalization of production and data exploitation. In this paper, a specific case of an automotive factory digital transformation is presented, including the description of a roadmap, an architecture, and a first set of analyses regarding quality, reliability and predictive maintenance in the Paint Shop. Data gathering, data visualization and statistical control are also commented. Some of the success stories include reduction of defect rates, optimization of processes and early detection of deviation of parameters, by using Artificial Intelligence algorithms on datasets with thousands of variables and hundreds of thousands of produced cars. These actions have meant savings for the factory, as well as productivity and efficiency improvements, showing that it is possible to obtain short-term victories while implementing a long-term tool. This methodology is currently being transferred to the rest of the factory, including stamping, body in white and assembly line.","PeriodicalId":215489,"journal":{"name":"2020 Eighth International Symposium on Computing and Networking (CANDAR)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123959841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An In-Network Parameter Aggregation using DPDK for Multi-GPU Deep Learning 基于DPDK的多gpu深度学习网络参数聚合

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00021

Masaki Furukawa, Tomoya Itsubo, Hiroki Matsutani

{"title":"An In-Network Parameter Aggregation using DPDK for Multi-GPU Deep Learning","authors":"Masaki Furukawa, Tomoya Itsubo, Hiroki Matsutani","doi":"10.1109/CANDAR51075.2020.00021","DOIUrl":"https://doi.org/10.1109/CANDAR51075.2020.00021","url":null,"abstract":"In distributed deep neural network using remote GPU nodes, communication occurs iteratively between remote nodes for gradient aggregation. This communication latency limits the benefit of distributed training with faster GPUs. In distributed deep learning using the remote GPUs, workload of gradient aggregation is imposed on a host machine. In this paper, we therefore propose to offload the gradient aggregation to a DPDK (Data Plane Development Kit) based network switch between the host machine and remote GPUs. In this approach, the aggregation process is completed in the network using extra computation resources in the network switch. We evaluate the proposed switch when GPUs and the host communicate with a standard IP communication and a PCI Express (PCIe) over 40Gbit Ethernet (40GbE) product, respectively. The evaluation results using a standard IP communication show that the aggregation is accelerated by 2.2-2.5x compared to the aggregation executed by a host machine. The results using the PCIe over 40GbE product show that the proposed switch outperforms the aggregation done by the host machine by 1.16x. This approach is thus useful for distributed training with multiple GPUs.","PeriodicalId":215489,"journal":{"name":"2020 Eighth International Symposium on Computing and Networking (CANDAR)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133473767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 3

[Title page i] [标题页i]

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/candar51075.2020.00001

引用次数: 0

Maximum Expressive Number of Threshold Neural Networks 阈值神经网络的最大表达数

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00023

Kenta Inoue

引用次数: 0

Efficient GPU Implementation for Solving the Maximum Independent Set Problem 求解最大独立集问题的高效GPU实现

2020 Eighth International Symposium on Computing and Networking (CANDAR) Pub Date : 2020-11-01 DOI: 10.1109/CANDAR51075.2020.00012

Tomohiro Imanaga, K. Nakano, Masaki Tao, Ryota Yasudo, Yasuaki Ito, Yuya Kawamata, Ryota Katsuki, Yusuke Tabata, Takashi Yazane, Kenichiro Hamano

{"title":"Efficient GPU Implementation for Solving the Maximum Independent Set Problem","authors":"Tomohiro Imanaga, K. Nakano, Masaki Tao, Ryota Yasudo, Yasuaki Ito, Yuya Kawamata, Ryota Katsuki, Yusuke Tabata, Takashi Yazane, Kenichiro Hamano","doi":"10.1109/CANDAR51075.2020.00012","DOIUrl":"https://doi.org/10.1109/CANDAR51075.2020.00012","url":null,"abstract":"An independent set of a graph is a subset of the nodes such that no two nodes in it are adjacent. The maximum independent set (MIS) problem is an optimization problem to find an independent set with the largest possible number of nodes. Since the MIS problem is NP-hard, finding optimal solutions for large graphs is quite difficult and approximation or heuristic algorithms are used to find independent sets as large as possible. The main contribution of this paper is to present an annealing-based parallel heuristic algorithm for finding approximate solutions for the MIS problem. We have implemented it to run on a computing server with multiple GPUs. The performance of our GPU implementation is compared with the other approaches including a GPU QUBO solver, Gurobi optimizer running on an Intel multicore server, and D-wave 2000Q quantum annealer. The experimental results for random graphs show that our GPU implementation for the MIS problem can find optimal or better solutions than the other approaches. In particular, for small random graphs with 128 nodes or less, our GPU implementation can find optimal solutions of the MIS problem in less than 20µs, while the other approaches take more than 10ms. Further, for large regular random graphs with 32M nodes, our GPU implementation can obtain 22.6%-38.5% larger independent sets. Thus, our parallel GPU implementation for solving the MIS problem is a potent approach to solve the MIS problem.","PeriodicalId":215489,"journal":{"name":"2020 Eighth International Symposium on Computing and Networking (CANDAR)","volume":"157 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115681818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 2