计算机应用 ›› 2013, Vol. 33 ›› Issue (09): 2428-2431.DOI: 10.11772/j.issn.1001-9081.2013.09.2428

• 先进计算 • 上一篇    下一篇

层流扩散燃烧在GPU上的并行计算和数值分析

魏浩洋1,2,3,曾国荪1,2,3,丁春玲4   

  1. 1. 国家高性能计算机工程技术中心 同济分中心,上海 200092;
    2. 同济大学 计算机科学与技术系,上海 200092;
    3. 国家高性能计算机工程技术中心 同济分中心,上海 200092;
    4. 同济大学 化学系,上海 200092
  • 收稿日期:2013-03-20 修回日期:2013-04-26 出版日期:2013-09-01 发布日期:2013-10-18
  • 通讯作者: 魏浩洋
  • 作者简介:魏浩洋(1988-),男,浙江温州人,硕士研究生,主要研究方向:并行分布处理、云计算;
    曾国荪(1964-),男,江西新干人,教授,博士生导师,博士,主要研究方向:并行计算、可信软件、信息安全;
    丁春玲(1965-),女,江西宜春人,高级工程师,硕士,主要研究方向:建模分析。
  • 基金资助:

    国家863计划项目;国家自然基金资助项目;上海市优秀学科带头人计划项目;教育部网络时代的科技论文快速共享专项研究课题

Parallel computing and numerical analysis of laminar diffusion combustion on GPU

WEI Haoyang1,2,ZENG Guosun1,2,DING Chunling3   

  1. 1. Department of Computer Science and Technology, Tongji University, Shanghai 200092, China;
    2. Tongji Branch, National Engineering and Technology Center of High Performance Computer, Shanghai 200092, China;
    3. Department of Chemistry, Tongji University, Shanghai 200092, China
  • Received:2013-03-20 Revised:2013-04-26 Online:2013-10-18 Published:2013-09-01
  • Contact: WEI Haoyang

摘要: 在实际工程应用中,使用传统的CPU串行计算来开展燃烧数值模拟往往难以满足对模拟速度的要求。利用GPU比CPU更强的计算能力,通过在交错网格上将燃烧物理方程离散化,使用预处理稳定双共轭梯度法(PBiCGSTAB)求解离散化方程,并且探索面向GPU编程的矩阵向量乘并行算法和逆矩阵向量乘并行算法,从而给出一种在GPU上数值求解层流扩散燃烧的可行方法。实验结果表明,GPU并行程序获得了相对串行CPU程序约10倍以上的加速效果,且计算结果与实际情况相符,因而所提方法是可行且高效的。

关键词: 层流扩散燃烧, 数值模拟, GPU并行计算, 预处理稳定双共轭梯度法, 加速比

Abstract: In practical engineering applications, using traditional CPU serial computation for combustion numerical simulation can hardly meet the requirements on simulation speed. This paper took the advantage of GPU which has more computing capability than CPU, by discretizing the combustion physical equations on staggered grid, solving the discrete equations with preconditioned bi-conjugate gradient stabilized (PBiCGSTAB) method, exploring the parallel algorithm of GPU-oriented matrix vector multiplication and the parallel algorithm of inverse matrix vector multiplication. Hence a feasible method for numerical calculation of laminar diffusion combustion on GPU was offered. The experimental results indicate that the parallel program on GPU achieved more than 10 times speedup relative to serial program on CPU. Since the calculated results are in line with the actual situation, the method is feasible and efficient.

Key words: laminar diffusion combustion, numerical simulation, GPU parallel computing, preconditioned bi-conjugate gradient stabilized method, speedup

中图分类号: