加入收藏夹
联系我们
关于本站
个人主页
西电导航
西安电子科技大学
    当前位置:首页>>学术讲座
On MapReduce Acceleration in Multi-GPU systems
时间:2016-05-04 08:43    点击:   所属单位:网络与信息安全学院
讲座名称 On MapReduce Acceleration in Multi-GPU systems
讲座时间 2016-05-05 15:30:00
讲座地点 北校区新科技楼 1012会议室
讲座人 Kuan-Ching Li
讲座人介绍

李冠憬教授是英工程科技学会 (IET) 国电与电气工程师学会 (IEEE) 高級会员,目前是台湾静宜大教授、兼任校長特別助理和国际岸事务处副主任。此外,李教授还担任內地多所大的客座、特聘与讲座教授。李教授荣获项奖励內外研究目,积极参与举办国际讨会,作主席、程序委员会主席等职务组织算科学与工程相会议国际期刊IJCSEIJESIJHPCN的主编与个国际期刊的副主委。其主要研究域包括分散式算、云计算、GPU算和平行程式设计,在此域已表多篇SCI/SCIE/EI期刊与国际会议论文、编辑了十多本技术专书国际出版社  Springer、  McGrawHill   Taylor&Francis 出版。

讲座内容 The potential of Big Data is widely recognized and organizations are collecting huge amounts of data. Most of them have the same goal: to extract "value" through sophisticated analysis for findings and decisions. Rapid advancements in computer and networking technologies have been critical to the IT revolution, driving the rapid decline in the cost and fast increase in the processing power of digital technologies. Concurrently, in the ever-changing world of software development, it’s extremely important to keep up with current technologies, methodologies and trends. MapReduce is a programming model introduced by Google for largescale data processing. Several studies have implemented MapReduce model on Graphic Processing Unit (GPU). However, most of them are based on the single GPU and bounded by GPU memory with inefficient atomic operations. In this presentation, we present  a standalone MapReduce system to utilize multiple GPUs, handle large-scale data processing beyond GPU memory limit, and eliminate serial atomic operations. Experimental results have demonstrated MGMR's effectiveness and promising the handling of large data sets.
转载请注明出处:西安电子科技大学学术信息网
如果您有学术信息或学术动态,欢迎投稿。我们将在第一时间确认并收录,投稿邮箱: meeting@xidian.edu.cn
Copyright © 2011-2017 西安电子科技大学 
开发维护:电子工程学院网络信息中心  管理员:meeting@xidian.edu.cn 站长统计: