One of our major developments over the last two years was the multi-threading or parallelization of the 3D programs. Users of the software often got results in minutes as opposed to hours using dual quad core machines. This was especially true for the boundary element method where the process of generating the system matrix and right hand side of the governing integral equations lends itself ideally to parallel processing. In addition, the dense matrix resulting from the boundary element method could be multi-threaded as well. Of course, what sounds simple to do was in fact quite challenging as the complete data structure of parallel algorithms is far more demanding than simple serial computing.

This development will now be the basis for implementing the same data structures and program implementation for the 2D programs. Although the benefit of parallel computation will perhaps not be as significant as experienced by 3D users, the speed improvement will certainly be a major plus. The speed increase will initially be found with the boundary element solver. However, some of these methods will eventually be used to the finite element software for mesh and matrix generation. We look forward to this development in V9.1.