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
复制标题

基于CPU和GPU集群的大规模并行化宇宙流体模拟

DOI:
10.1007/978-3-642-53962-6_18
复制
发表时间:
2013-05
期刊:
Compfluid
影响因子:
--
通讯作者:
Zhu Weishan
Zhu Weishan
中科院分区:
其他
文献类型:
--
作者:
Wang Long;Cao Zongyan;Feng Longlong;Zhu Weishan

文献摘要

参考文献

被引文献

相似文献

我们提出了基于CPU和GPU集群的大规模三维超音速流体宇宙学模拟的并行实现。我们的开发是基于一个名为wigeon的CPU代码。结果表明,与原始的顺序Fortran代码相比,在单个GPU上可以获得19-31的加速比(取决于特定的GPU卡)。此外,我们的结果表明,纯MPI并行化可以很好地扩展到10000个CPU核。此外,还提出了一种CPU/GPU混合并行化方案,并对不同数量的CPU/GPU单元(受计算资源限制,最多256块GPU卡)的加速比和伸缩性进行了详细的分析。我们的高可伸缩性和加速比依赖于区域分解方法、算法的优化和一系列技术来优化CUDA实现,特别是在GPU上的存储器访问模式。我们相信,这种混合MPI+CUDA代码可以成为10Peta级及更高级别计算的优秀候选者。
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 GPU. We believe this hybrid MPI + CUDA code can be an excellent candidate for 10 Peta-scale computing and beyond.
DOI: 10.1145/1513895.1513905
发表时间: 2009-03
影响因子: 3
作者:
P. Micikevicius
通讯作者: P. Micikevicius
DOI: 10.1115/1.3601282
发表时间: 1968-09
期刊: Journal of Applied Mechanics
影响因子: --
作者:
G. Batchelor;A. D. Young
通讯作者: G. Batchelor;A. D. Young
DOI: 10.1006/jcph.1996.0130
发表时间: 1996-06-01
影响因子: 4.1
作者:
Jiang, GS;Shu, CW
通讯作者: Shu, CW
DOI: 10.1006/jcph.2000.6443
发表时间: 2000-05-20
影响因子: 4.1
作者:
Balsara, DS;Shu, CW
通讯作者: Shu, CW
DOI: 10.1080/10618560802238275
发表时间: 2008-08
影响因子: 1.3
作者:
J. Tölke;M. Krafczyk
通讯作者: J. Tölke;M. Krafczyk