[petsc-users] Matrix assembly too slow

[Karl Rupp]

Hi,

> 3) If this is the expected behavior, and it takes so long for a 28000 x
> 28000 linear system, it would be impossible to scale up to millions of
> dofs. It seems hard to believe that the cost communicating the matrix
> with matAssemblyBegin-End is much bigger or even comparable to the cost
> of actually calculating the values with numerical integration.

You should also keep in mind that since you are assembling a dense 
matrix, you have *much* higher memory requirements as compared to the 
frequently used H-matrix approaches for BEM. You really need to exploit 
the structure of the integral kernel 'to scale up to millions of dofs'.

Best regards,
Karli