an insightful program performance tuning chain for gpu computing

ISCAS OpenIR

	an insightful program performance tuning chain for gpu computing
	Jia Haipeng; Zhang Yunquan; Long Guoping; Yan Shengen
	2012
会议名称	12th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2012
会议录名称	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
页码	502-516
会议日期	September 4, 2012 - September 7, 2012
会议地点	Fukuoka, Japan
收录类别	EI
ISSN	0302-9743
ISBN	9783642330773
部门归属	(1) Lab. of Parallel Software and Computational Science Institute of Software Chinese Academy of Sciences China; (2) College of Information Science and Engineering Ocean University of China China; (3) State Key Laboratory of Computing Science Chinese Academy of Sciences China; (4) Graduate University of Chinese Academy of Sciences China
摘要	It is challenging to optimize GPU kernels because this progress requires deep technical knowledge of the underlying hardware. Modern GPU architectures are becoming more and more diversified, which further exacerbates the already difficult problem of performance optimization. This paper presents an insightful performance tuning chain for GPUs. The goal is to help non-expert programmers with limited knowledge of GPU architectures implement high performance GPU kernels directly. We achieve it by providing performance information to identify GPU program performance bottlenecks and decide which optimization methods should be adopted, so as to facilitate the best match between algorithm features and underlying hardware characteristics. To demonstrate the usage of tuning chain, we optimize three representative GPU kernels with different compute intensity: Matrix Transpose, Laplace Transform and Integral on both NVIDIA and AMD GPUs. Experimental results demonstrate that under the guidance of our tuning chain, performance of those kernels achieves 7.8~42.4 times speedup compared to their nai¨ve implementations on both NVIDIA and AMD GPU platforms. © 2012 Springer-Verlag.; It is challenging to optimize GPU kernels because this progress requires deep technical knowledge of the underlying hardware. Modern GPU architectures are becoming more and more diversified, which further exacerbates the already difficult problem of performance optimization. This paper presents an insightful performance tuning chain for GPUs. The goal is to help non-expert programmers with limited knowledge of GPU architectures implement high performance GPU kernels directly. We achieve it by providing performance information to identify GPU program performance bottlenecks and decide which optimization methods should be adopted, so as to facilitate the best match between algorithm features and underlying hardware characteristics. To demonstrate the usage of tuning chain, we optimize three representative GPU kernels with different compute intensity: Matrix Transpose, Laplace Transform and Integral on both NVIDIA and AMD GPUs. Experimental results demonstrate that under the guidance of our tuning chain, performance of those kernels achieves 7.8~42.4 times speedup compared to their nai¨ve implementations on both NVIDIA and AMD GPU platforms. © 2012 Springer-Verlag.
关键词	Hardware Laplace Transforms Optimization Program Processors
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/15798
专题	中国科学院软件研究所
推荐引用方式 GB/T 7714	Jia Haipeng,Zhang Yunquan,Long Guoping,et al. an insightful program performance tuning chain for gpu computing[C],2012:502-516.