Rice University

Events at Rice

Seminar

Electrical and Computer Engineering
Dean of Engineering
Faculty Host: Lin Zhong

Speaker: Zheng (Eddy) Zhang
Assistant Professor, Computer Science
Rutgers University

ECE Seminar Series: ORION: A Framework for GPU Occupancy Autotuning and Resource Allocation

Tuesday, May 9, 2017
4:00 PM  to 5:00 PM

1064  Duncan Hall
Rice University
6100 Main St
Houston, Texas, USA

The data-parallel GPU architecture has become increasingly popular today because of its power efficiency and programmability. An important feature of modern GPU architecture is tunable occupancy. Occupancy measures the number of active threads on a GPU with respect to single thread resource demand and hardware scheduling capacity. High occupancy allows a large number of threads to run simultaneously and hide memory latency from each other, yet may also increase resource contention and power consumption. Occupancy tuning is a challenge problem. A program running at different occupancy levels can have a 3X difference in running time. However, there has been limited exploration toward occupancy-aware GPU program optimization. We introduce ORION – an automatic GPU occupancy tuning and resource allocation framework. ORION generates binary code for any given occupancy level with (near-)optimal resource al- location and is capable of finding the best occupancy level with static iterative compilation and dynamic runtime adaptation. We demonstrate the efficacy of ORION by autotuning twelve representative benchmarks from the Rodinia benchmark suite and CUDA SDK on two different GPU architectures, and obtaining up to 1.57 times speedup, 68.5% memory resource saving, and 28.5% power saving with the baseline of highly optimized code compiled by nvcc

Biography of Zheng (Eddy) Zhang:
Zheng (Eddy) Zhang is an assistant professor in the Department of Computer Science at Rutgers University. She received the B.S. degree in Electronic Engineering from Shanghai Jiao Tong University, in 2004, and the M.S. and Ph.D. degrees in Computer Science from the College of William & Mary, in 2007 and 2012, respectively. Her research is generally in the area of compilers and programming systems, with a focus on development, compilation, and execution of large-scale irregular application on many-core parallel architecture. She is the recipient of Google Faculty Research Award 2014, the Best Paper Award at PPoPP 2010, and the Best Student Paper Award at QEST 2008.



<<   May 2017   >>
S M T W T F S
1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30 31

Search for Events


Quicklinks