one of the reason of slow performance in FEM analysis is construction of elastic stiffness matrix with loops.
There is a large for loops with size of the number of elements. This can quickly add significant overhead when NE is large since each line in the loop will be interpreted in each iteration. Vectorization should be applied for the sake of efficiency.
Dear Ali
There are some files as well as a link, as attachment. It is hoped that these sources would be useful.
best regards
Panji
http://onlinelibrary.wiley.com/doi/10.1029/94WR02269/abstract
Dear Ali
There are some files as well as a link, as attachment. It is hoped that these sources would be useful.
best regards
Panji
http://onlinelibrary.wiley.com/doi/10.1029/94WR02269/abstract
Hi,
A similar thing was done here with edge elements
http://arxiv.org/pdf/1409.4618.pdf
but I have successfully applied a similar approach for nodal elements in Matlab. You can send me a mail if you want details.
br.
Antti
The elements loop can be parallelised very simply using OpenMP or MPI for multiple cores. Vectorisation on modern processors can be carried out for the maths operations within the loop and is largely implemented in maths libraries. Some source code is available at http://parafem.org.uk. This is described in the text book Programming the Finite Element Method.
http://parafem.org.uk
http://www.amazon.co.uk/gp/product/1119973341?keywords=programming%20the%20finite%20element%20method&qid=1447431907&ref_=sr_1_1&sr=8-1
- Badges
- Science topic