Large-scale parallelization based on CPU and GPU cluster for cosmological fluid simulations

	Large-scale parallelization based on CPU and GPU cluster for cosmological fluid simulations
	Meng, Chen 1,2; Wang, Long 1; Cao, Zongyan 1,3; Feng, Long-long4 ; Zhu, Weishan 4
刊名	COMPUTERS & FLUIDS
	2015-03-30
卷号	110 页码:152-158
关键词	Cosmological hydrodynamics WENO GPU Hierarchical memory Heterogeneous Large-scale
英文摘要	We present our parallel implementation for large-scale cosmological simulations of 3D supersonic fluids based on CPU and GPU clusters. Our developments are based on a CPU code named WIGEON. It is shown that, compared to the original sequential Fortran code, a speedup of 19-31 (depending on the specific GPU card) can be achieved on single GPU. Furthermore, our results show that the pure MPI parallelization scales very well up to 10 thousand CPU cores. In addition, a hybrid CPU/GPU parallelization scheme is introduced and a detailed analysis of the speedup and the scaling on the different number of CPU/GPU units are presented (up to 256 GPU cards due to computing resource limitation). Our high scalability and speedup rely on the domain decomposition approach, optimization of the algorithm and a series of techniques to optimize the CUDA implementation, especially in the memory access pattern on CPU. We believe this hybrid MPI + CUDA code can be an excellent candidate for 10 Peta-scale computing and beyond. (C) 2014 Elsevier Ltd. All rights reserved.
学科主题	天文和天体物理
WOS标题词	Science & Technology ; Technology
类目[WOS]	Computer Science, Interdisciplinary Applications ; Mechanics
研究领域[WOS]	Computer Science ; Mechanics
关键词[WOS]	SCHEMES
收录类别	SCI
语种	英语
WOS记录号	WOS:000350535100016
内容类型	期刊论文
源URL	[http://libir.pmo.ac.cn/handle/332002/14928]
专题	紫金山天文台_南极天文中心
作者单位	1.Chinese Acad Sci, Supercomp Ctr, Comp Network Informat Ctr, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China 3.Chinese Acad Sci, Natl Astron Observ, Beijing 100012, Peoples R China 4.Chinese Acad Sci, Purple Mt Observ, Nanjing 210008, Jiangsu, Peoples R China
推荐引用方式 GB/T 7714	Meng, Chen,Wang, Long,Cao, Zongyan,et al. Large-scale parallelization based on CPU and GPU cluster for cosmological fluid simulations[J]. COMPUTERS & FLUIDS,2015,110:152-158.
APA	Meng, Chen,Wang, Long,Cao, Zongyan,Feng, Long-long,&Zhu, Weishan.(2015).Large-scale parallelization based on CPU and GPU cluster for cosmological fluid simulations.COMPUTERS & FLUIDS,110,152-158.
MLA	Meng, Chen,et al."Large-scale parallelization based on CPU and GPU cluster for cosmological fluid simulations".COMPUTERS & FLUIDS 110(2015):152-158.