[petsc-users] MatMatSolve for MUMPS disappeared?

[Alexander Grayver]

On 18.09.2012 20:14, Hong Zhang wrote:
> MatMatSolve_MUMPS()  is never been supported.
> It seems mumps support multiple rhs now.
> If you need it, we can implement MatSolve_MUMPS().

Hi Hong,

Thanks for reply.

Is MatSolve in any way more efficient than KSPSolve?
As far as I understand for both cases one ends up with a loop over RHS,
but KSPSolve allows to use iterative solvers (I don't need them at the 
moment though).
I would then use latter option and do this loop myself.

>
>     There was one issue with it for sequential mode that has been
>     fixed here:
>     http://petsc.cs.iit.edu/petsc/releases/petsc-3.3/rev/8badc49a596e
>
>
> This change prevents  calling of MatMatSolve_Basic().
> How many rhs vectors (or number of columns in your rhs matrix)?
> MUMPS only supports centralized rhs.  Scattering many rhs vectors to
> a sequential dense matrix is non-scalable.

That is right. The request on that has recently appeared in the mumps 
mailing list and they replied there are no plans to develop it.
I have usually < 10^3 and this is not a bottleneck, so I'm fine with a 
"inefficient" loop over RHS.

But, for those who are chasing performance and have many sprase RHS 
(e.g., this is the case when one needs to solve many adjoint problems 
with delta function as a RHS) I can say that using MUMPS' multiple RHS 
mode is a way more efficient than solving them in a loop. This is 
probably related to some internals where MUMPS is able to take advantage 
of getting many RHS at once.

-- 
Regards,
Alexander

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20120919/f31d7327/attachment.html>