My own hardware was absolutely cutting-edge in 2005. As it stands, I can do linear analyses without too much trouble, non-linear takes a while, and fully dynamic non-linear / multiphysics brings my system to a halt. This is on hardware that was acquired specifically for the purpose, so I can't make any assumptions about what anyone can run on their desktops. In any case, the huge matrix equations that comprise FE analyses generally can not be reduced, other than by analyzing a coarser mesh. With my software, there are many ways I can potentially reduce the number of calculations without affecting accuracy, since I am only removing superfluous data sets from consideration.
-Sean
On Nov 17, 2009, Alan James <alanjames@xtra.co.nz> wrote:
Sean,Thanks you're doing an awesome work there.My thought is that if the programs a bit slow crunching figures at the moment, that won't necessarily be the case in afew years time. Computers are always getting faster & more powerful. Also I was looking at the article on the sub that gotcertification through FEA results & their computer was taking 30 minutes for non linear buckling analyses.Alan