The OpenSees Community

Dear Experts!

I am quite well in simulating with OpenSees (at least RC structures). However, I am working on large-detailed 3D models:

1441 Nodes, 504 Nonlinear BeamColumn Elements and 1296 ShellMITC4 Elements. Some larger models also exist!

I am curious about the SPEED of analysis in Parallel OpenSeesSP and OpenSeesMP. I have checked some of the lectures on parallel Computing in OpenSees and now my questions:

1) Please, give me an example to feel the advantage of shifting from OpenSees to OpenSeesSP/OpenSeesMP! (Speed of Analysis)

2) Does OpenSeesSP need knowledge of Parallel Computing/Parallel Machines? or it just needs some modifications in solver options, numberer and ...
3) In the lecture notes, I have seen some infromation regarding OpenSeesSP for Windows. Is there any for Linux since we have some powerfull computing-core in our institute which are running only on linux!

4) Does OpenSeesMP need knowledge of Parallel Computing (I think yes!)?
5) Does OpenSeesMP exist for linux?

I am sorry if I asked lots of questions,
Thank you in advance.

A.H.A

1) if you do not have 2 many constraints you should see a spepdup of between 2 and 8.
2) no .. just change the solver type to Mumps
3) we release binaries for windows and mac, not for linux. most parallel usage is on linux clusters. hoever, given that the
configurations are all different, we cannot make a release for them all. help is provided in example Makefile's that have been used
to build on different machines.
4) it requires more knowledge than 1 .. the potential sppedup is greater.
4) see 3 above.

Dear Dr. fmk,

Thanks for your completely clear response.

A.H.A