[petsc-dev] [issue1595] Issues of limited number of MPI communicators when having many instances of hypre boomerAMG with Moose

[Matthew Knepley]

On Tue, Apr 3, 2018 at 6:02 PM, Rob Falgout hypre Tracker <
hypre-support at llnl.gov> wrote:

>
> Rob Falgout <rfalgout at llnl.gov> added the comment:
>
> In my opinion, if the user is doing something simultaneously with a hypre
> solve (not the typical case, I think), then they should be responsible for
> ensuring that the communicators don't conflict.  It will also be easier to
> control the number of communicators used in those situations where there
> are limits.
>

That seems like a solipsistic way to operate, and it does not scale very
well. Substitute "other library" for "user" in
what you said. Now all libraries but Hypre must protect themselves from
Hypre messages. Why is Hypre special?
Doesn't it seem like you guys should segregate your messages into a
different space?

   Matt


> Just my 2 cents.
>
> -Rob
>
> ____________________________________________
> hypre Issue Tracker <hypre-support at llnl.gov>
> <http://cascb1.llnl.gov/hypre/issue1595>
> ____________________________________________
>



-- 
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

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20180403/5379c5b4/attachment.html>