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
Handling Large-scale Partitionable Workloads on Heterogeneous Networked Compute Platforms
时间:2018-06-22 17:55    点击:   所属单位:计算机学院
讲座名称 Handling Large-scale Partitionable Workloads on Heterogeneous Networked Compute Platforms
讲座时间 2018-06-29 09:00:00
讲座地点 北校区主楼四区107会议室
讲座人 Bharadwaj Veeravalli教授(新加坡国立大学)
讲座人介绍 Bharadwaj Veeravalli, Senior MIEEE, MIEEE-CS, is currently with the Department of Electrical and Computer Engineering at the National University of Singapore (NUS), Singapore, as a tenured Associate Professor. He was awarded a gold medal for best PhD thesis from IISc Bangalore, India in 1994. From 1994-1997, he worked as  Post-Doctoral Fellow at the Concordia University, Montreal, Quebec, Canada and at Canadian Aerospace Corporation(1997-1998), His major research interests include Cloud/Grid/Cluster Computing, Scheduling in Parallel and Distributed Systems, Embedded Computing, Bioinformatics & Computational Biology, and Multimedia Computing. He is one of the earliest researchers in Divisible Load Theory (DLT). He had secured and directed several externally funded large-scale projects in his research areas and published over 85 journal papers and 90 high quality Conference papers. He has co-authored three research monographs in the areas of PDS, Distributed Databases, Networked Multimedia Systems and, energy optimization for multiprocessor platforms, in 1996, 2003, 2005 and, 2018, respectively. He is currently serving the Editorial Board of IEEE Transactions on Parallel & Distributed Systems and had served IEEE TC, IEEE Transactions on Cloud Computing, IEEE Transactions on SMC-A, and Cluster Computing, as an Associate Editor. He has served as a lead guest editor for special issues on Cloud Computing themes with IEEE Transactions on Computers (2014) and IEEE Transactions on Cloud Computing (2015). He was a visiting professor position with Hunan University, China, and has served as a TPC member in several International Conferences and also as Conference Co-chair. More information can be found in http://cnl-ece.nus.edu.sg/elebv/.
讲座内容 With growing interest and ability in handling Big/Exa-scale Data on Cloud and networked-compute platforms,unless servicing provisioning is efficient would result in system inefficiency. An elegant way to handle large-scale compute loads is by efficient resource management by dynamically allocating the compute resources as and when required. However for very large-scale loads graceful resource scaling would not significantly gain in terms of monetary cost. Alternatively, one can adopt designing efficient load distribution/allocation strategies that could result in significant gain in terms of time and cost performances, as one can use software-tunable parameters to optimize. In this talk, I will introduce and highlight the benefits of adopting multi-round strategies for handling large-scale partitionable workloads that demand homogeneous processing. This means entire workload can divided and redistributed among the available compute nodes to minimize the overall time performance. The designed strategies offer flexibility in deciding the sizes of the loads to be allocated to compute nodes, thus becoming a viable alternative solution to current day Map-Reduce paradigm. I will share my recent past experiences in designing such efficient strategies and highlight certain application examples in image processing and multimedia retrieval to demonstrate the efficiency of the proposed techniques.
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 © 2011-2018 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: