On Thu, May 24, 2012 at 2:16 PM, Mark F. Adams <mark.adams at columbia.edu>wrote:
> Is Eisenstat really worth it? > > Given a memory centric future and that Jed, even now, is not seeing a win > .. maybe with a regular grid and R/B ordering you can skip a whole pass > through the data but for AIJ (I assume that is what ex5 uses) lexagraphic > ordering is probably a better model and its does not look like a big win to > me. > Maybe Eisenstat should split the matrix storage similar to the factorization kernels? > > Mark > > On May 24, 2012, at 2:48 PM, Barry Smith wrote: > > > > > On May 24, 2012, at 1:20 PM, Jed Brown wrote: > > > >> On Wed, May 23, 2012 at 2:52 PM, Jed Brown <jedbrown at mcs.anl.gov> > wrote: > >> On Wed, May 23, 2012 at 2:26 PM, Barry Smith <bsmith at mcs.anl.gov> > wrote: > >> > >> Note that you could use -pc_type eisenstat perhaps in this case > instead. Might save lots of flops? I've often wondered about doing Mark's > favorite chebyshev smoother with Eisenstat, seems like it should be a good > match. > >> > >> [0]PETSC ERROR: --------------------- Error Message > ------------------------------------ > >> [0]PETSC ERROR: No support for this operation for this object type! > >> [0]PETSC ERROR: Cannot have different mat and pmat! > >> > >> Also, I'm having trouble getting Eisenstat to be more than very > marginally faster than SOR. > > > > There is still a bug related to Eisenstat. If you are trying to use > it with kspest you are likely hitting the bug. > > > > I will fix the bug when I have time. > > > > Barry > > > >> > >> > >> I think we should later be getting the eigenvalue estimate by applying > the preconditioned operator to a few random vectors, then orthogonalizing. > The basic algorithm is to generate a random matrix X (say 5 or 10 columns), > compute > >> > >> Y = (P^{-1} A)^q X > >> > >> where q is 1 or 2 or 3, then compute > >> > >> Q R = Y > >> > >> and compute the largest singular value of the small matrix R. The > orthogonalization can be done in one reduction and all the MatMults can be > done together. Whenever we manage to implement a MatMMult and PCMApply or > whatever (names inspired by VecMDot), this will provide a very low > communication way to get the eigenvalue estimates. > >> > >> > >> I want to turn off norms in Chebyshev by default (they are very > wasteful), but how should I make -mg_levels_ksp_monitor turn them back on? > I'm already tired of typing -mg_levels_ksp_norm_type unpreconditioned. > > > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120524/8560357a/attachment.html>