北京清华大学高精尖研究中心
分享到

Initial Steps toward Making GPU a First-Class Computing Resource: Sharing and Resource Management

Date: 2017-08-07
Writer:

Initial Steps toward Making GPU a First-Class Computing Resource:  Sharing and Resource Management

I. Details of the lecture 

Topic: Initial Steps toward Making GPU a First-Class Computing Resource:  Sharing and Resource Management

Time: 10:00am, Aug 8, 2017

Location: Room 8-206, ROHM Building,Tsinghua University

Speaker: Jun Yang   Professor, Dept. of  Electrical  and Computer Engineering, University of Pittsburgh, PA, USA

II. Introduction to the speaker 

Jun Yang is a Professor of Electrical and Computer Engineering Department at the University of Pittsburgh. She received her Ph.D. from The University of Arizona, 2002, and became an assistant professor at UC Riverside (2002-2006) prior to joining University of Pittsburgh. Jun’s research is in the broad area of computer architecture and her recent focuses include GPU designs, emerging memory technologies, interconnection networks, 3D integration, and power and thermal management techniques. Jun is a recipient of NSF CAREER award in 2008, IEEE MICRO Top Picks award in 2010, and best paper awards of ISLPED 2013 and ICCD 2007.

III. Content of lecture 

GPUs have evolved from traditional graphics accelerators into core compute engines for a broad class of general-purpose applications. However, current commercial offerings fall short of the great potential of GPUs largely because they cannot be managed as easily as the CPU. The enormous amount of hardware resources are often greatly underutilized as there is hardly effective architectural support for fully managing them even when the GPU is shared across multiple applications. 

To make GPUs a first-class controllable resource, we developed new architecture features to enable fine-grained sharing of GPUs, termed Simultaneous Multi-kernel (SMK), in a similar way the CPU achieves sharing via simultaneous multithreading (SMT). With SMK, different applications can co-exist in every streaming multiprocessor of a GPU. High resource utilization can be achieved by exploiting heterogeneity of different application behaviors. Resource apportion among sharers are developed for fairness, throughput, quality-of-services, etc. We also envision that SMK can enable better manageability of GPUs and new features such as more efficient synchronization mechanisms within an application.

Hot News / Related recommendation
2017 - 02 - 28
点击次数: 0
The Center hosted the high-end forum under the theme of “Future Chip 2016: Challenges and Opportunities in Design Automation” at Tsinghua University Information Science and Technology Building on Dece...
2017 - 02 - 28
点击次数: 0
I. Details of the lecture Topic: Optimization of Non-Volatile FPGA Time: Wednesday, December 14, 2016, 12:10 - 13:10Venue: Room 10-206, Rohm Building, Tsinghua UniversitySpeaker: Hu Jingtong...
2017 - 02 - 28
点击次数: 0
i. Seminar InformationTopic: Flash performance and reliability optimization: from chip, controller to systemTime: Wednesday, December 14, 2016, 12:10 - 13:10Location: Room 10-206, Rohm ...
2017 - 02 - 28
点击次数: 0
i. Seminar InformationTopic: The Design of A User-Centric Mobile SystemTime: 2016 December 14(Wed)13:10 - 13:40Location: Room 10-206, Rohm Building, Tsinghua University Speaker: Dr. Pi-Cheng Hsiu...
Center Address
Address: Beijing Innovation Center for Future Chip, Building A, 3# Heqing Road, Haidian District, Beijing, China
Telephone: 010-62788711
WeChat: THU-ICFC
北京清华大学高精尖研究中心
Copyright ©2005 - 2013 BEIJING INNOVATION CENTER FOR FUTURE CHIP
犀牛云提供云计算服务
Focus on WeChat public number
Focus on WeChat public number
Focus on WeChat public number