[petsc-users] MatSetValue - Faster if guaranteed local processor?

[Ganesh Vijayakumar]

Hello,

On Fri, May 8, 2015 at 5:24 PM, Jed Brown <jed at jedbrown.org> wrote:

> Ganesh Vijayakumar <ganesh.iitm at gmail.com> writes:
>
> > Hello,
> >
> >    Is there any way I can get MatSetValue to run faster if I can
> guarantee
> > that the entries are local to that processor?
>
> Use MatSetValuesBlocked if relevant.  Determining on/off process is not
> a bottleneck.  Profile carefully before assuming where the time is
> spent.
>

Thanks. I have a sparse matrix. The current access order doesn't allow easy
blocking of the values being set.

Thanks for the advice on profiling too. I am working towards it. I was
initially working on getting it working.

ganesh


-- 
ganesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150508/4f88cb81/attachment.html>