[petsc-users] Compile flags for a Blue Gene\Q

[Jed Brown]

Please always use "reply-all" so that your messages go to the list.
This is standard mailing list etiquette.  It is important to preserve
threading for people who find this discussion later and so that we do
not waste our time re-answering the same questions that have already
been answered in private side-conversations.  You'll likely get an
answer faster that way too.

Christian Klettner <ucemckl at ucl.ac.uk> writes:

> Dear Jed,
> We have been testing your suggestion and found that we can not use ESSL
> 5.1.1 which is installed on their system. From the installation page on
> the PETSc website it does suggest that ESSL is difficult to bind in. Are
> you using ESSL 5.1.1-0.beta to get around this issue? 

No, it was just what was available when I built that configuration.

> Do you see much improvement when going from a 'standard' BLAS to ESSL
> blas for sparse matrix multiplications?  

No, BLAS is not used in sparse matrix multiplication.  The BLAS
implementation usually makes very little difference since Level-1
operations are most of what is used by sparse iterative methods.  Sparse
direct solvers eventually produce large supernodes, so BLAS matters
there.