论文部分内容阅读
结合有限元法 ,研究了 EBE策略在热传导问题并行计算方面的应用 ,给出了瞬态热传导问题的隐式 EBE并行算法。根据所用 CPU个数并行计算单元刚度矩阵 ,而总刚度矩阵 K不需要组装 ,仅在“单元级”上进行各种计算。就模型实验规模 10 2 4阶而言 ,机器存储量比传统算法降低了 40 %左右 ,在拥有共享内存的并行计算机 Challenge上的数值实验表明 ,问题的规模越大 ,并行效率越高 ,所给算法非常有效
Combined with the finite element method, the application of EBE strategy in parallel computation of heat conduction problems is studied. An implicit EBE parallel algorithm for transient heat conduction problems is given. The element stiffness matrix is calculated in parallel based on the number of CPUs used, whereas the total stiffness matrix K does not need to be assembled, and various calculations are performed only at “element level.” For the model experiment scale of 10 2 4, the machine storage is reduced by about 40% compared with the traditional algorithm. Numerical experiments on a parallel computer Challenge with shared memory show that the larger the scale of the problem, the higher the parallel efficiency The algorithm is very effective