dolfin team mailing list archive

[Nuno David Lopes <ndl@xxxxxxxxxxxxxx>]

> And using -ksp_type gmres -pc_type hypre -pc_hypre_type boomeramg is
> failing for solving with A? 
isn't it the same as using dolfin
____________________________________________________
LinearSolver solver(gmres,amg);
____________________________________________________

> This doesn't make sense to me, I regularly 
> use it for this problem.  Is the memory blowup occurring during PC
> assembly?
I think so, it blows when the output is:
____________________________________________________
Solving linear system of size 566769 x 566769 (Krylov solver).
____________________________________________________ 
if we want to pass command line options we have to 
use:
----
 dolfin::SubSystemsManager::initPETSc(argc,argv);
----
but doesn't a 
---
KrylovSolver solver(gmres,sor) for instance 
---
override the options passed trhu (argc,argv) ? 
  
> Are you sure the matrix is correct?
I hope so...it works with other proconditioners (SOR, is the faster till now)
and i reduced the problem to the simple case 
> Is it parallel or  
> serial?
serial...
> How much storage does the matrix alone take?
+-1.5g 
> Does it work  
> correctly for smaller problem sizes?
It worked fine with a poisson problems with bigger systems
> Have you tried with ML (-pc_type 
> ml -mat_no_inode)?  Multigrid should work very well for this problem.
Don't have it... I will build PETSc with ml
>
> Jed

thanks.

-- 
Nuno David Lopes

e-mail:ndl@xxxxxxxxxxxxxx        (FCUL/CMAF)
           nlopes@xxxxxxxxxxxxxxx    (ISEL)
http://ptmat.ptmat.fc.ul.pt/%7Endl/ 

Thu Sep 25 10:35:19 WEST 2008