...
首页> 外文期刊>Physica, A. Statistical mechanics and its applications >Parallel performance and accuracy of lattice Boltzmann and traditional finite difference methods for solving the unsteady two-dimensional Burger's equation
【24h】

Parallel performance and accuracy of lattice Boltzmann and traditional finite difference methods for solving the unsteady two-dimensional Burger's equation

机译:求解非定常二维Burgers方程的格子Boltzmann与传统有限差分方法的并行性能和准确性

获取原文
获取原文并翻译 | 示例
           

摘要

Lattice Boltzmann methods are gaining recognition in the field of computational fluid dynamics due to their computational efficiency. In order to quantify the computational efficiency and accuracy of the lattice Boltzmann method, it is compared with efficient traditional finite difference methods such as the alternating direction implicit scheme. The lattice Boltzmann algorithm implemented in previous studies does not approach peak performance for simulations where the data involved in computation per time step is more than the cache size. Due to this, data is obtained from the main memory and this access is much slower than access to cache memory. Using a cache-optimized lattice Boltzmann algorithm, this paper takes into account the full computational strength of the lattice Boltzmann method. The coin parison is performed on both a single processor and multiple processors. (c) 2005 Elsevier B.V. All rights reserved.
机译:格子Boltzmann方法由于其计算效率而在计算流体动力学领域获得认可。为了量化格子Boltzmann方法的计算效率和准确性,将其与高效的传统有限差分方法(如交替方向隐式方案)进行了比较。在以前的研究中实施的晶格Boltzmann算法在仿真中没有达到峰值性能,因为每个时间步长所涉及的数据大于缓存大小。因此,从主存储器中获取数据,并且这种访问比对高速缓存存储器的访问要慢得多。使用缓存优化的格子Boltzmann算法,本文考虑了格子Boltzmann方法的全部计算强度。硬币型坯在单个处理器和多个处理器上执行。 (c)2005 Elsevier B.V.保留所有权利。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号