Thank you for your time
today. It was a very good discussion. Here's what I captured (please add/modify
what I may have missed):
Present: Leonid Meyerguz
(Microsoft), Rich Graham (ORNL), Richard Barrett (ORNL), Torsten Hoefler (ISU),
Alexander Supalov (Intel)
- Opens & introductions
- Scope of the effort
-
Rich
-
Minimum subset consistent with the rest of MPI, for performance/memory footprint
optimization
-
Danger of splitting MPI, hence against optional features in the
standard
-
Both blocking & nonblocking belong to the
core
-
Torsten
-
Some collectives may go into selectable
subsets
-
MPI_ANY_SOURCE considered harmful
-
Leonid
-
Flexible support for optional features, means for choosing and advertising level
of compliance/set of features
- See enclosed
email for Alexander's POV
- General discussion
snapshots
- Support of
subsets: some or all? If some, possible linkage problems in static apps (or dead
calls). If all, where's the gain?
- Optional:
really optional (may be not present) or selectable (are present but may be
unused)?
- Performance
penalty for unused subsets: implementation matter or standard
choice?
- Portability
may be limited to certain class of applications (think FT, master-slave
runs)
- All we design
needs to be implementable, complexity needs to be
controlled
- An ability to
use certain set of subsets should not preclude pulling in other modules if
necessary
- Whatever we
do, it should not conflict with the ABI
efforts
- Need to stay
nice and be nicer wrt to the libraries (think threading) and keep things
simple
- The
simplification argument, if put first, may not be liked by
some
- Next steps
- Please
comment on these minutes, and add/modify what I may have
missed
- I'll prepare
a couple of slides by next week summarizing our discussion so far; again, your
feedback will be most welcome
- At the
meeting, it may be great to meet F2F briefly and discuss any
eventual loose ends before the presentation at the Forum; I'll see to
this
Best regards.
Alexander
--
Dr
Alexander Supalov
Intel
GmbH
Hermuelheimer
Strasse 8a
50321
Bruehl, Germany
Phone: +49
2232 209034
Mobile:
+49 173 511 8735
Fax: +49 2232
209029
---------------------------------------------------------------------
Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen Germany
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer
Registergericht: Muenchen HRB 47456 Ust.-IdNr.
VAT Registration No.: DE129385895
Citibank Frankfurt (BLZ 502 109 00) 600119052
This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.