On 13.11.2012 14:11, Jack Poulson wrote: > I can't help but interject. Sparse direct solvers are not nonscalable > "in general". General purpose solvers have to worry about pivoting, > which greatly complicates the factorization and tends to get in the > way of scalability. Several sparse-direct solvers use one-dimensional > frontal distributions (e.g., MUMPS and SPOOLES) in order to simplify > pivoting, though MUMPS uses a two-dimensional distribution at the root > front. > > It is well-known how to make multifrontal Cholesky factorization > scalable (for example, see one of the many papers by > Gupta/Karypis/Kumar/Joshi from the 90's). Making multifrontal > triangular solves scalable is another matter.
Jack, Thanks for the comment. Could you comment on or give a reference to the blow up in memory one observes when running general purpose direct solvers in parallel? P.S. Does Clique support pivoting? -- Regards, Alexander
