Abstract:
Aiming at the problems such as large quantity of computing and low processing speed for grid subdivision module in pre-processing system, the sequential grid subdivision algorithm is parallelized, which is implemented by using multi-thread parallel mechanism. This parallel algorithm is also tested. Experimental results show this algorithm has absolute advantage to sequential algorithm on processing speed, and the accelerated speed and parallel efficiency increase as grid accuracy or the number of basic entities rises.
Key words:
pre-processing,
grid subdivision,
parallelizing,
multi-thread
摘要: 针对前处理中网格剖分模块计算量大、处理速度慢等问题,对网格剖分串行算法进行并行化处理,利用多线程并行机制加以实现。同时对该并行算法进行测试。实验结果表明,该算法相对于串行算法在处理速度上具有绝对优势,且其加速度和并行效率均会随网格精度的提高以及基本实体个数的增多而提高。
关键词:
前处理,
网格剖分,
并行化,
多线程
CLC Number:
CHEN Ru; ZHANG Wen-yao; NING Jian-guo. Parallelizing Technique of Finite-difference Grid Subdivision in 3D Pre-processing[J]. Computer Engineering, 2009, 35(4): 13-15.
陈 茹;张文耀;宁建国. 三维前处理有限差分网格剖分的并行化技术[J]. 计算机工程, 2009, 35(4): 13-15.