ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Thu, Sep 28, 2006 at 05:48:53PM +0200, Anders Logg wrote:
> On Thu, Sep 28, 2006 at 05:42:11PM +0200, Garth N. Wells wrote:
> > Anders Logg wrote:
> > > On Thu, Sep 28, 2006 at 03:48:30PM +0200, Garth N. Wells wrote:
> > >> Could we add something to the FFC output to indicate the
> > >> presence/absence of interior and boundary integrals? Then we can skip
> > >> the interiors/boundary assembly loop in DOLFIN if there are no
> > >> interior/boundary terms.
> > >>
> > >> Garth
> > > 
> > > Yes, I have wanted to do this before, but never gotten around to doing
> > > it. It's a simple fix that I can fix quickly.
> > > 
> > > This would take care of the BLAS format for cases when there are no
> > > boundary contributions?
> > >
> > 
> > It should do. I do however recall some problems with the Intel compiler
> > when not supplying a function for the boundary integral, but this could
> > be fixed with a compiler flag.
> > 
> > Garth
> 
> I'll just add a function that answers true or false on whether or not
> there is a boundary contribution. The function is supplied (but empty)
> even if there is no contribution.
> 
> /Anders
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev

Should be fixed now in both FFC and DOLFIN. Check if it works.

The logic in the assemble functions is getting a little to
complicated: three forms a, L and M are assembled simultaneously and
each form may or may not be a null pointer and if the pointer is
nonzero then there may or may not be an interior contribution and
there may or may not be a boundary contribution.

/Anders