[petsc-dev] Ghost update in VecAssembly?

[Matthew Knepley]

On Tue, Mar 1, 2016 at 7:33 AM, Jed Brown <jed at jedbrown.org> wrote:

> Matthew Knepley <knepley at gmail.com> writes:
>
> > It was suggested to me that we add VecGhostUpdate() to VecAssembly().
> This
> > sounds
> > reasonable to me. What do you think?
>
> Uh, why?  What is the most common use of VecAssembly?  Seems like
> building b or x_0 to solve A x = b, in which case you don't need updated
>

This is a good argument.


> ghost values.  Also, who uses VecGhost _and_ VecAssembly for the same
> Vec?  That's quite wasteful; if you have local indexing, why not just
> use VecGhostUpdate(...,SCATTER_REVERSE)?
>

I do not understand this argument. If you form the local part of each vector
using VecGetArray(), then you must call both VecAssembly() and
VecGhostUpdate(),
which I think is the complaint.

   Matt

-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20160301/890d3909/attachment.html>