Large scale satellite imagery simulations with physically based ray tracing on tianhe-1A supercomputer

ISCAS OpenIR

	Large scale satellite imagery simulations with physically based ray tracing on tianhe-1A supercomputer
	Wu, Changmao (1); Zhang, Yunquan (2); Yang, Congli (1); Zhang, Y.(changmaowu@gmail.com)
	2014
会议名称	15th IEEE International Conference on High Performance Computing and Communications, HPCC 2013 and 11th IEEE/IFIP International Conference on Embedded and Ubiquitous Computing, EUC 2013
页码	549-556
会议日期	November 13, 2013 - November 15, 2013
会议地点	Zhangjiajie, Hunan, China
收录类别	EI
出版地	IEEE Computer Society
ISBN	9780769550886
部门归属	(1) Institute of Software, CAS, China; (2) State Key Laboratory of Computer Architecture, Institute of Computing Technology, CAS, China; (3) University of Chinese Academy of Sciences, Beijing, China
摘要	Developing highly scalable algorithms for satellite imagery simulations is becoming increasingly important as scientists inquire to understand the mechanism of satellite imagery before satellites are launched into orbit. Although physically based ray tracing technique for image rendering has produced some of the most realistic images to date, studies on satellite imagery simulations using this technique are still very less to be seen, due in large part to both the complex physical processes and the computational difficulties of the mathematical models. In this paper, we present a highly scalable physically based ray tracer for satellite imagery simulations. Our ray tracer is based on a Master-Worker-Receiver framework which can overcome the performance bottleneck of Master node. Besides, a novel sample distribution strategy is presented by the authors, aiming at removing high additional computation overhead which is introduced by the currently available pixel distribution strategy. Compared to the pixel distribution strategy, our sample distribution strategy drops the computation overhead by 0.25 to 4 times. We also discuss the issue with granularity of assignment partitioning for Inter-Nodes and Intra-Nodes, then a hybrid scheduling strategy combining static and dynamic scheduling strategies is presented. Experiments show that our physically based ray tracer almost reaches to a linear speedup by using 16,800 CPU cores on Tianhe-1A Supercomputer. Our ray tracer is more efficient and highly scalable. © 2013 IEEE.; Developing highly scalable algorithms for satellite imagery simulations is becoming increasingly important as scientists inquire to understand the mechanism of satellite imagery before satellites are launched into orbit. Although physically based ray tracing technique for image rendering has produced some of the most realistic images to date, studies on satellite imagery simulations using this technique are still very less to be seen, due in large part to both the complex physical processes and the computational difficulties of the mathematical models. In this paper, we present a highly scalable physically based ray tracer for satellite imagery simulations. Our ray tracer is based on a Master-Worker-Receiver framework which can overcome the performance bottleneck of Master node. Besides, a novel sample distribution strategy is presented by the authors, aiming at removing high additional computation overhead which is introduced by the currently available pixel distribution strategy. Compared to the pixel distribution strategy, our sample distribution strategy drops the computation overhead by 0.25 to 4 times. We also discuss the issue with granularity of assignment partitioning for Inter-Nodes and Intra-Nodes, then a hybrid scheduling strategy combining static and dynamic scheduling strategies is presented. Experiments show that our physically based ray tracer almost reaches to a linear speedup by using 16,800 CPU cores on Tianhe-1A Supercomputer. Our ray tracer is more efficient and highly scalable. © 2013 IEEE.
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16637
专题	中国科学院软件研究所
通讯作者	Zhang, Y.(changmaowu@gmail.com)
推荐引用方式 GB/T 7714	Wu, Changmao ,Zhang, Yunquan ,Yang, Congli ,et al. Large scale satellite imagery simulations with physically based ray tracing on tianhe-1A supercomputer[C]. IEEE Computer Society,2014:549-556.