itaps-parallel Reminder: Phone Conference Wednesday Oct 31 at 9am PDT

[Devine, Karen D.]

Reminder:  We will have a subcommittee phone conference on Wednesday Oct 31
at 9am PDT.  Lori will host using the usual phone numbers.

Here is a tentative agenda.  We'll stick to big picture issues, rather than
deal with frightening syntax or ghostly spiritual issues.

-  Having just caught up on the email discussions, I have to thank Carl for
doing a great job of putting the discussions into a consistent framework.
In the "no good deed goes unpunished" spirit, I'd like Carl to begin the
discussions by summarizing his model.  Then I'd like Onkar, Tim and Vitus to
say whether or not Carl's description supports the functionality they need
and whether their subsections of the interface (submitted for draft zero)
are, or can be made to be, consistent with that description.
HOMEWORK:  Read Carl's latest email and be prepared to say how it fits with
your subsection of draft zero.

-  A discussion of parts versus processors.  In my experience, it is easier
for both application and tool developer to deal with only parts/partitions
as much as possible, rather than present both parts and processors.  Clearly
some mapping is needed, but I think it is good to have operations defined as
much as possible on parts and partitions only.  Both Carl and Onkar alluded
to this idea as well in their separate processor and ProcPart layers.  So we
will discuss whether we should tie "global" operations to a partition rather
than an MPI communicator.  We may cover this ground in Carl's discussion
above; if so we'll move on to....

-  Uses that we need to support, and what exactly is a root set?   I can
think of several uses that we might want to support.  The easiest is a
single mesh with a single partition of it; in this case, what is in the root
set?  Is there one root set per part?  Or one per processor?  Another case
is having a single mesh with multiple partitions of it (e.g., crash
simulations where the elements are partitioned one way and the surfaces are
partitioned differently).  Does the root set change?  Does our parallel
model support multiple "Partition Instances"?  Third, suppose we have
multiple meshes that are partitioned as if they were one mesh (e.g., a
multiphysics simulation where one partition is applied to multiple meshes).
What does the root set contain in this case?   Would we have a single
"Partition Instance" with multiple Mesh Instances?  Should the relationship
between Partitions and Meshes be strictly hierarchical?   Fourth, suppose we
have multiple meshes, each with their own partition.  Am I making you crazy
yet?

If we get through these three points in one phone conference, you are
permitted to celebrate and break into the trick-or-treat candy early.  If
there are other hairy, scary, big-picture issues we should discuss, please
email them to me.  Thanks.

Karen