Dear all,
after a couple of years developing with Fenics on MacOS X, I stumbled upon
two dreary problems: it won't provide binaries anymore, and it does not
play well with multithreading. I was then thinking about switching to
deal.II, but I hope you can help me in dispelling some doubts.
My analyses are non-linear and parametric, hence my problems with Fenics: I
parallelized single analyses, but I found many hidden race conditions.
So, here are my doubts about moving to deal.II:
1 - Is it truly thread-safe? As in running a single analysis, or running
several (no data sharing), without problems.
2 - Are there facilities of some kind that can help in generating the
assembly code? In Fenics I just specified the weak formulations.
3 - Can I output some kind of time-dependant file, like a PVD? I see VTK,
but no PVD files.
4 - I see Trilinos is included, is it possible to avoid completely
PETSc/SLEPc, since they're not AFAIK thread-safe?
5 - Is there a multithreaded linear solver and vector operations, or should
I write them myself?
6 - Can I use my MLK library without recompiling deal.II?
7 - Is eliminating binary distributions (DMGs and such) on the horizon?
Thank you very much for your help on any of these points!
Franco
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
