The PETSc interface for MatMatSolve() for SuperLU_Dist does directly use the
multiple right hand side solver of SuperLU so perhaps you can use this as it
will be faster than one solve at a time.
As Hong mentioned, MUMPS still requires centralizing the entire right hand
side to one process
Matt, Barry & Hong,
Thanks a lot for response.
I may be wrong, but I think that people chose direct solvers for their
robustness and ability to solve for many RHS very quickly.
Surely I don't expect my solution to be sparse and it is not, but at
least in electromagnetics it is pretty common to h
Hi Alexander,
I am just curious how other sparse direct solver (eg. PaStiX) works on
your problem. Right now, I only have single RHS, but I will need
multi-RHS soon. For my single RHS problem, PaStiX works better for my
sparse matrix pattern (helmholtz like). If you tried PaStiX, I would
be happy
Hi Barry,
Thanks for answer. I should have asked about that from the very
beginning actually.
I get 2 times decrease in performance with 20 RHS, I can imagine how
slow it will be when I will use thousands of them,
moreover it will require a lot of memory to store them as a dense matrix.
On 11.1
Reading MUMPS manual, mumps only supports centralized dense and sparse rhs,
while solution can be centralized or distributed, but only in the
dense format (because manual
does not mention sparse solu).
It would be easy to add MatMatSolve() in Petsc-mumps interface for dense rhs,
but will require v
Alexander,
Do you expect that solving with a sparse right hand side will result in an
answer that is also sparse? Normally once you've done the lower triangular
solve and then the upper triangular solve won't the result be dense?
Barry
On Dec 12, 2011, at 3:10 AM, Alexander Grayver
On Mon, Dec 12, 2011 at 3:10 AM, Alexander Grayver
wrote:
> Hi Barry,
>
> Thanks for answer. I should have asked about that from the very beginning
> actually.
> I get 2 times decrease in performance with 20 RHS, I can imagine how slow
> it will be when I will use thousands of them,
> moreover it
Hello,
I used to use MUMPS directly with sparse multiple RHS. Now I use MUMPS
through PETSc interface and solution for multiple RHS takes 1.5-2 times
longer (MatMatSolve).
My first question is whether do you use multiple RHS internally or you
solve one-by-one?
Second guess concerns the option:
On Dec 11, 2011, at 9:29 AM, Alexander Grayver wrote:
> Hello,
>
> I used to use MUMPS directly with sparse multiple RHS. Now I use MUMPS
> through PETSc interface and solution for multiple RHS takes 1.5-2 times
> longer (MatMatSolve).
> My first question is whether do you use multiple RHS int